MIxed Method Design PDF

04-Creswell (Designing)-45025.
qxd 5/16/2006 8:35 PM Page 58
CHAPTER 4
CHOOSING A MIXED
METHODS DESIGN
R esearch designs are procedures for collecting, analyzing, interpreting,

and reporting data in research studies. They represent different mod-
els for doing research, and these models have distinct names and
procedures associated with them. Rigorous research designs are important
because they guide the methods decisions that researchers must make dur-
ing their studies and set the logic by which they make interpretations at the
end of studies. Once a researcher has selected a mixed methods approach for
a study, the next step is to decide on the specific design that best addresses
the research problem. What designs are available, and how do researchers
decide which one is appropriate for their studies? Mixed methods researchers
need to be acquainted with the major types of mixed methods designs and
the common variants among these designs. Important considerations when
choosing designs are knowing the intent, the procedures, and the strengths
and challenges associated with each design. Researchers also need to be
familiar with the timing, weighting, and mixing decisions that are made in
each of the different mixed methods designs.
This chapter will address
The classifications of designs in the literature

The four major types of mixed methods designs, including their
intent, key procedures, common variants, and inherent strengths and
challenges
58
04-Creswell (Designing)-45025.qxd 5/16/2006 8:35 PM Page 59
Choosing a Mixed Methods Design 59
Factors such as timing, weighting, and mixing, which influence the

choice of an appropriate design
CLASSIFICATIONS OF MIXED METHODS DESIGNS

Researchers benefit from being familiar with the numerous classifications of
mixed methods designs found in the literature. These classifications repre-
sent different disciplines, and they use different terminology. Researchers
should be aware of the range of mixed methods design types, as well as the
discipline-based discussions of mixed methods designs.
Methodologists writing about mixed methods research have devoted a
great deal of attention to classifying the different types of mixed methods
designs. In the final chapter of the Handbook of Mixed Methods in Social
and Behavioral Research, Tashakkori and Teddlie (2003b) noted that they
had found nearly 40 different types of mixed methods designs in the litera-
ture. Creswell, Plano Clark, et al. (2003) have summarized the range of these
classifications. Their summary has been updated, and a list of 12 classifica-
tions is included in Table 4.1. These classifications represent diverse social
science disciplines, including evaluation, health research, and educational
research, which span the past 15 years of scholarly writings about mixed
methods approaches. The different types and various classifications speak to
the evolving nature of mixed methods research.
Seeing the long list of design types in Table 4.1 may be overwhelming.
It is easy to get lost in the details, as these classifications are drawn from
different disciplines, have emphasized different facets of mixed methods
designs, and lack consistency in the names of the designs. It may even
appear that little agreement exists among these authors and that there are
an infinite number of design options. In fact, although authors have empha-
sized different features and used different names, there are actually more
similarities than differences among these classifications. Based on these
similarities, we feel that a parsimonious and functional classification can be
created. Thus we advance four major mixed methods designs, with variants
within each type.
THE FOUR MAJOR TYPES OF MIXED METHODS DESIGNS

The four major types of mixed methods designs are the Triangulation Design,
the Embedded Design, the Explanatory Design, and the Exploratory Design.
The following sections provide an overview of each of these designs: their
use, procedures, common variants, and challenges.
60 DESIGNING AND CONDUCTING MIXED METHODS RESEARCH
Table 4.1 Mixed Method Design Classifications
Author Mixed Method Designs Discipline
Greene, Caracelli, Initiation Evaluation

and Graham Expansion
(1989) Development
Complementary
Triangulation
Patton (1990) Experimental design, qualitative data, and Evaluation

content analysis
Experimental design, qualitative data, and
statistical analysis
Naturalistic inquiry, qualitative data, and
Naturalistic inquiry, quantitative data, and
Morse (1991) Simultaneous triangulation Nursing

QUAL + quan
QUAN + qual
Sequential triangulation
QUAL quan
QUAN qual
Steckler, McLeroy, Model 1: Qualitative methods to develop Public health education

Goodman, Bird, quantitative measures
and McCormick Model 2: Qualitative methods to explain
(1992) quantitative findings
Model 3: Quantitative methods to embellish
qualitative findings
Model 4: Qualitative and quantitative
methods used equally and parallel
Greene and Component designs Evaluation

Caracelli (1997) Triangulation
Complementary
Expansion
Integrated designs
Iterative
Embedded or nested
Holistic
Transformative

Morgan (1998) Complementary designs Health research
Qualitative preliminary
Quantitative preliminary
Qualitative follow-up
Quantitative follow-up
Tashakkori and Mixed method designs Educational research
Teddlie (1998) Equivalent status (sequential or parallel)
Dominantless dominant (sequential or
parallel)
Multilevel use
Mixed model designs:
I. Confirmatory, qualitative data,
statistical analysis, and inference
II. Confirmatory, qualitative data,
qualitative analysis, and inference
III. Exploratory, quantitative data,
IV. Exploratory, qualitative data,
V. Confirmatory, quantitative data,
VI. Exploratory, quantitative data,
VII. Parallel mixed model
VIII. Sequential mixed model
Creswell (1999) Convergence model Educational policy
Sequential model
Instrument-building model
Sandelowski Sequential Nursing
(2000) Concurrent
Iterative
Sandwich
Creswell, Plano Sequential explanatory Educational research
Clark, Gutmann, Sequential exploratory
and Hanson Sequential transformative
(2003) Concurrent triangulation
Concurrent nested
Concurrent transformative
(Continued)
Table 4.1 (Continued)

Creswell, Fetters, Instrument design model Primary medical care
and Ivankova Triangulation design model
(2004) Data transformation design model
Tashakkori and Multistrand designs Social and behavioral
Teddlie (2003b) Concurrent mixed designs research
Concurrent mixed method design
Concurrent mixed model design
Sequential mixed designs
Sequential mixed method design
Sequential mixed model design
Multistrand conversion mixed designs
Multistrand conversion mixed method design
Multistrand conversion mixed model design
Fully integrated mixed model design
SOURCE: Adapted from Creswell, Plano Clark, et al. (2003, pp. 216-217, Table 8.1).
The Triangulation Design
The most common and well-known approach to mixing methods is the

Triangulation Design (Figure 4.1a) (Creswell, Plano Clark, et al., 2003). The pur-
pose of this design is to obtain different but complementary data on the same
topic (Morse, 1991, p. 122) to best understand the research problem. The intent
in using this design is to bring together the differing strengths and nonoverlap-
ping weaknesses of quantitative methods (large sample size, trends, generaliza-
tion) with those of qualitative methods (small N, details, in depth) (Patton, 1990).
This design and its underlying purpose of converging different methods has
been discussed extensively in the literature (e.g., Jick, 1979; Brewer & Hunter,
1989; Greene et al., 1989; Morse, 1991). This design is used when a researcher
wants to directly compare and contrast quantitative statistical results with quali-
tative findings or to validate or expand quantitative results with qualitative data.
Triangulation Design Procedures. The Triangulation Design is a one-phase

design in which researchers implement the quantitative and qualitative
(a) Triangulation Design
QUAN QUAL
Interpretation based on
QUAN + QUAL results
(b) Triangulation Design: Convergence Model
QUAN QUAN
QUAN
data data
results
collection analysis
Compare
Interpretation
and
QUAN + QUAL
contrast
QUAL QUAL
QUAL
data data
results
collection analysis
(c) Triangulation Design: Data Transformation Model (Transforming QUAL data into QUAN)
QUAN
data QUAN data analysis
collection
Compare and
Interpretation
interrelate two
QUAN + QUAL
QUAN data sets
QUAL QUAL Transform

data data QUAL into
collection analysis quan data
(d) Triangulation Design: Validating Quantitative Data Model
QUAN data QUAN

QUAN
collection: data
results
Survey analysis
Validate QUAN
Interpretation
results with
QUAN + qual
qual results
qual
qual
data collection: qual
data
Open-ended results
analysis
survey items
Figure 4.1 (Continued)

(e) Triangulation Design: Multilevel Model
Level 1:
QUAN
data collection, analysis, results
Level 2:
QUAL Overall
data collection, analysis, results interpretation
Level 3:
QUAN
data collection, analysis, results
Figure 4.1 The Triangulation Design
methods during the same timeframe and with equal weight (see Figure 4.1a).
The single-phase timing of this design is the reason it has also been referred
to as the concurrent triangulation design (Creswell, Plano Clark, et al.,
2003). It generally involves the concurrent, but separate, collection and
analysis of quantitative and qualitative data so that the researcher may best
understand the research problem. The researcher attempts to merge the two
data sets, typically by bringing the separate results together in the interpre-
tation or by transforming data to facilitate integrating the two data types dur-
ing the analysis. Jenkins (2001) single-phase study (appendix A) of rural
adolescent perceptions of alcohol and other drug resistance is an example of
a Triangulation Design. She collected and analyzed quantitative and qualita-
tive data and merged the two data sets into one overall interpretation, in
which she related the quantitative results to the qualitative findings.
Variants of the Triangulation Design. The four variants are the convergence
model, the data transformation model, the validating quantitative data
model, and the multilevel model. The first two models differ in terms of how
the researcher attempts to merge the two data types (either during interpre-
tation or during analysis), the third model is used to enhance findings from
a survey, and the fourth is used to investigate different levels of analysis.
The convergence model (Figure 4.1b) represents the traditional model
of a mixed methods triangulation design (Creswell, 1999). In this model, the
researcher collects and analyzes quantitative and qualitative data separately
on the same phenomenon and then the different results are converged (by
comparing and contrasting the different results) during the interpretation.
Researchers use this model when they want to compare results or to validate,
confirm, or corroborate quantitative results with qualitative findings. The

purpose of this model is to end up with valid and well-substantiated conclu-
sions about a single phenomenon. For example, Anderson, Newell, and
Kilcoyne (1999) converged their quantitative survey results with their quali-
tative focus group findings to better understand the motivations of college
student plasma donors.
Researchers may choose to use the data transformation model (Figure
4.1c) (Creswell et al., 2004). This model also involves the separate collection
and analysis of quantitative and qualitative data sets. However, after the ini-
tial analysis, the researcher uses procedures to transform one data type into
the other data type. This is accomplished by either quantifying qualitative
findings or qualifying quantitative results (Tashakkori & Teddlie, 1998). This
transformation allows the data to be mixed during the analysis stage and facil-
itates the comparison, interrelation, and further analysis of the two data sets.
The study of parental values by Pagano, Hirsch, Deutsch, and McAdams
(2002) is an example of this model. In their study, they derived qualitative
themes from the qualitative data and then scored those themes dichoto-
mously as present or not present for each participant. These quantified
scores were then analyzed with the quantitative data, using correlations and
logistical regression to identify relationships between categories, as well as
gender and racial differences.
Researchers use the validating quantitative data model (Figure 4.1d)
when they want to validate and expand on the quantitative findings from a
survey by including a few open-ended qualitative questions. In this model,
the researcher collects both types of data within one survey instrument.
Because the qualitative items are an add-on to a quantitative survey, the items
generally do not result in a rigorous qualitative data set. However, they pro-
vide the researcher with interesting quotes that can be used to validate and
embellish the quantitative survey findings. For example, Webb, Sweet, and
Pretty (2002) included qualitative questions in addition to their quantitative
survey measures in their study of the emotional and psychological impact of
mass casualty incidents on forensic odontologists. Webb et al. used the qual-
itative data to validate the quantitative results from the survey items.
The fourth variant of the Triangulation Design is what Tashakkori and
Teddlie (1998) referred to as multilevel research (p. 48). In a multilevel
model (Figure 4.1e), different methods (quantitative and qualitative) are
used to address different levels within a system. The findings from each level
are merged together into one overall interpretation. For example, Elliott and
Williams (2002) studied an employee counseling service using qualitative
data at the client level, qualitative data at the counselor level, qualitative data
with the directors, and quantitative data for the organizational level.
Strengths of the Triangulation Design. This design has a number of strengths

and advantages, including the following:
The design makes intuitive sense. Researchers new to mixed methods

often choose this design. It was the design first discussed in the liter-
ature (Jick, 1979), and it has become a framework for thinking about
mixed methods research.
It is an efficient design, in which both types of data are collected
during one phase of the research at roughly the same time.
Each type of data can be collected and analyzed separately and inde-
pendently, using the techniques traditionally associated with each
data type. This lends itself to team research, in which the team can
include individuals with both quantitative and qualitative expertise.
Challenges in Using the Triangulation Design. Although this design is the

most popular mixed methods design, it is also probably the most challenging
of the four major types of designs. Here are some of the challenges facing
researchers using a variant of the Triangulation Design as well as options for
addressing them.
For all variants:
Much effort and expertise is required, particularly because of the con-

current data collection and the fact that equal weight is usually given
to each data type. This can be addressed by forming a research team
that includes members who have quantitative and qualitative exper-
tise, by including researchers who have quantitative and qualitative
expertise on graduate committees, or by training single researchers in
both quantitative and qualitative research.
Researchers may face the question of what to do if the quantitative
and qualitative results do not agree. These differences can be difficult
to resolve and may require the collection of additional data. The ques-
tion then develops as to what type of additional data to collect, quan-
titative data, qualitative data, or both? Chapter 6 discusses the
collection of additional data or the reexamination of existing data to
address this challenge.
For the convergence model:
Researchers need to consider the consequences of having different

samples and different sample sizes when converging the two data sets.
Different sample sizes are inherent in the design because quantitative
and qualitative data are usually collected for different purposes

(generalization vs. in-depth description, respectively). Researchers
can consider collecting large qualitative samples or weighting the
cases (see Chapter 6).
It can be very challenging to converge (integrate) two sets of very dif-
ferent data and their results in a meaningful way. Chapter 7 provides
techniques for building comparison matrices. In addition, researchers
need to design their studies so that the quantitative and qualitative data
address the same concepts. This strategy facilitates merging the data.
For the data transformation model:
Researchers need to develop procedures for transforming data and

make decisions about how the data will be transformed. In general, it
is easier for researchers to quantify their qualitative data by transform-
ing qualitative codes or themes into counts or ratings (see Chapter 7).
The Embedded Design
The Embedded Design is a mixed methods design in which one data set
provides a supportive, secondary role in a study based primarily on the other
data type (see Figure 4.2a) (Creswell, Plano Clark, et al., 2003). The premises
of this design are that a single data set is not sufficient, that different ques-
tions need to be answered, and that each type of question requires different
types of data. Researchers use this design when they need to include quali-
tative or quantitative data to answer a research question within a largely
quantitative or qualitative study. This design is particularly useful when a
researcher needs to embed a qualitative component within a quantitative
design, as in the case of an experimental or correlational design. In the exper-
imental example, the investigator includes qualitative data for several rea-
sons, such as to develop a treatment, to examine the process of an intervention
or the mechanisms that relate variables, or to follow up on the results of an
experiment.
Embedded Design Procedures. The Embedded Design mixes the different

data sets at the design level, with one type of data being embedded within a
methodology framed by the other data type (Caracelli & Greene, 1997). For
example, a researcher could embed qualitative data within a quantitative
methodology, as might be done in an experimental design, or quantitative
data could be embedded within a qualitative methodology, as could be done
(a) Embedded Design
QUAN Interpretation QUAL Interpretation

based on based on
or
qual QUAN(qual) quan QUAL(quan)
results results
(b) Embedded Design: Embedded Experimental Model
qual qual
QUAN Intervention QUAN
before after
premeasure postmeasure
intervention intervention
Interpretation
based on
QUAN(qual)
results
qual
during
intervention
(c) Embedded Design: Embedded Correlational Model
QUAN
Predictors
QUAN QUAN
Predictors Outcome Interpretation
based on
QUAN QUAN(qual)
Predictors results
qual
process
Figure 4.2 The Embedded Design
in a phenomenology design (see Figure 4.2a). The Embedded Design

includes the collection of both quantitative and qualitative data, but one of
the data types plays a supplemental role within the overall design. An
Embedded Design can use either a one-phase or a two-phase approach for
the embedded data (see Figure 4.2b), and the quantitative and qualitative
data are used to answer different research questions within the study
(Hanson et al., 2005). For example, Rogers et al. (2003) (appendix B) embed-
ded qualitative data within their experimental design in two different ways:
before the intervention, to inform the development of the treatment, and
after the intervention, to explain the treatment results.
It can be a challenge to differentiate between a study using an Embedded
Design and a study using one of the other mixed methods designs. The key
question is whether the secondary data type is playing a supplemental role
within a design based on the other data type. Consider the question: Would
the results of the secondary data type be useful or meaningful if they were
not embedded within the other data set? For example, Rogers et al.s (2003)
qualitative explanation of the treatment results would not make much sense
or have value if there had not been outcomes measured from an experimen-
tal study using those treatments.
Variants of the Embedded Design. Although many variants of the Embedded

Design are possible, there are two variants that we will discuss. These are the
experimental model and the correlational model.
The embedded experimental model (Figure 4.2b) may be the most com-
monly used variant of the Embedded Design (Creswell, Fetters, & Plano
Clark, 2005). This model is defined by having qualitative data embedded
within an experimental design (such as a true experiment or a quasiexperi-
ment). The priority of this model is established by the quantitative, experi-
mental methodology, and the qualitative dataset is subservient within that
methodology. This design can either be used as a one-phase or a two-phase
approach, in which the timing reflects the purpose for including the qualita-
tive data (Creswell et al., 2005; Sandelowski, 1996). For example, in a one-
phase approach, qualitative data can be embedded during the intervention
phase (see Figure 4.2b), such as when a researcher wants to qualitatively
examine the process of the intervention in addition to the quantitative out-
comes. This model has also been referred to as a concurrent nested mixed
methods design (Creswell, Plano Clark, et al., 2003). Alternatively, qualitative
data can come before or after the intervention in a two-phase model (see
Figure 4.2b). These sequential approaches are useful when a researcher
needs qualitative information before the intervention, to shape the interven-
tion, to develop an instrument, or to select participants, or after the inter-
vention, to explain the results of the intervention or to follow up on the
experiences of participants with certain types of outcomes. Victor, Ross, and
Axford (2004) collected qualitative data (diaries and audiotapes of treatment
sessions) to examine participant perspectives during their intervention trial
of a health promotion intervention for people with osteoarthritis of the knee.
Sports psychologists Evans and Hardy (2002a, 2002b) followed up on the

results of their experimental study of a goal-setting intervention for injured
athletes by interviewing participants from each of the treatment groups to
better interpret the findings of the intervention study.
The correlational model (Figure 4.2c) is another embedded variant, in
which qualitative data are embedded within a quantitative design. In this
design, researchers collect qualitative data as part of their correlational study
to help explain how the mechanisms work in the correlational model. For
example, Aikens (2004) is conducting a study of the factors relating depres-
sion and diabetes as moderated by race. Within his larger correlational study,
he is embedding qualitative interviews about beliefs and experiences with
depression for African American patients with diabetes to help explain the
predictive relationships.
Strengths of the Embedded Design. The advantages specific to this design

include the following:
It can be used when a researcher does not have sufficient time or

resources to commit to extensive quantitative and qualitative data
collection because one data type is given less priority than the other.
This design may be logistically more manageable for graduate students
because one method requires less data than the other method.
This design may be appealing to funding agencies because the
primary focus of the design is traditionally quantitative, such as an
experiment or a correlational analysis.
Challenges in Using the Embedded Design. There are many challenges

associated with the variants of the Embedded Design. These challenges, and
suggested strategies for dealing with them, include the following.
For all variants:
The researcher must specify the purpose of collecting qualitative (or

quantitative) data as part of a larger quantitative (or qualitative) study.
Researchers can state these as the primary and secondary purposes
for the study. See Chapter 5 for further discussion of examples for
writing these primary and secondary purpose statements.
It can be difficult to integrate the results when the two methods are
used to answer different research questions. However, unlike the
Triangulation Design, the intent of the Embedded Design is not to
converge two different data sets collected to answer the same ques-
tion. Researchers using an Embedded Design can keep the two sets
of results separate in their reports or even report them in separate

papers (see Chapter 8 for further discussion about these writing
strategies).
Few examples exist and little has been written about embedding quan-
titative data within traditionally qualitative designs. Researchers may
consider the timing of the quantitative data in relation to the larger
qualitative design, as suggested by Sandelowski (1996) and Creswell
et al. (2005) for experimental designs with embedded qualitative data.
For the embedded experimental model:
The researcher must decide at what point in the experimental study

to collect the qualitative data (before, during, or after the interven-
tion). This decision should be made based on the intent for including
the qualitative data (e.g., to shape the intervention, to explain the
process of participants during treatment, or to follow up on results of
the experimental trial).
For before-intervention approaches, the researcher needs to decide
which qualitative results will be used in the quantitative phase and
to consider how to plan the quantitative phase before the qualitative
phase has been conducted. Again, the qualitative data collection
should be carefully designed to match the intent for including qualita-
tive data, such as to develop an instrument or shape the intervention.
For during-intervention approaches, the qualitative data collection
may introduce potential treatment bias that affects the outcomes of
the experiment. Suggestions for addressing this bias through collect-
ing unobtrusive data are discussed in Chapter 6.
For after-intervention approaches, decisions must be made about
which aspects of the trial will be further explored, and the researcher
must specify the criteria used to select the participants for the follow-
up data collection. Researchers may want to follow up in depth only
with participants who received the treatment or with select cases
based on positive and negative treatment outcomes, as discussed fur-
ther in Chapter 6.
The Explanatory Design
The Explanatory Design is a two-phase mixed methods design (see

Figure 4.3a). The overall purpose of this design is that qualitative data helps
explain or build upon initial quantitative results (Creswell, Plano Clark, et al.,
2003). For example, this design is well suited to a study in which a researcher
needs qualitative data to explain significant (or nonsignificant) results, outlier
results, or surprising results (Morse, 1991). This design can also be used
when a researcher wants to form groups based on quantitative results and
follow up with the groups through subsequent qualitative research (Morgan,
1998; Tashakkori & Teddlie, 1998) or to use quantitative participant charac-
teristics to guide purposeful sampling for a qualitative phase (Creswell, Plano
Clark, et al., 2003).
Explanatory Design Procedures. The Explanatory Design (also known as the

Explanatory Sequential Design) is a two-phase mixed methods design (see
Figure 4.3a). This design starts with the collection and analysis of quantitative
data. This first phase is followed by the subsequent collection and analysis of
qualitative data. The second, qualitative phase of the study is designed so that
it follows from (or connects to) the results of the first quantitative phase.
Because this design begins quantitatively, investigators typically place greater
emphasis on the quantitative methods than the qualitative methods. Aldridge
et al.s (1999) study (appendix C) of classroom environments is an example
of an Explanatory Design. They started with a quantitative survey study and
identified statistically significant differences and anomalous results. They
then followed up these results with an in-depth qualitative study to explain
why these results occurred.
Variants of the Explanatory Design. There are two variants of the Explanatory
Design: the follow-up explanations model and the participant selection model.
Although both models have an initial quantitative phase followed by a quali-
tative phase, they differ in the connection of the two phases, with one focus-
ing on results to be examined in more detail and the other on the appropriate
participants to be selected (see center boxes of Figures 4.3b and 4.3c). They
also differ in the relative emphasis often placed on the two phases.
The follow-up explanations model (Figure 4.3b) is used when a
researcher needs qualitative data to explain or expand on quantitative results
(Creswell, Plano Clark, et al., 2003). In this model, the researcher identifies
specific quantitative findings that need additional explanation, such as statis-
tical differences among groups, individuals who scored at extreme levels, or
unexpected results. The researcher then collects qualitative data from partic-
ipants who can best help explain these findings. In this model, the primary
emphasis is usually on the quantitative aspects. Ivankovas (2004) disserta-
tion study of doctoral students persistence in an online learning environ-
ment is an example of this variant. In the initial quantitative phase, she
collected quantitative survey data to identify factors predictive of students
(a) Explanatory Design
04-Creswell (Designing)-45025.qxd
QUAN qual QUAN qual
results
5/16/2006
(b) Explanatory Design: Follow-up Explanations Model

8:35 PM
QUAN QUAN Identify qual qual

QUAN qual Interpretation
data data results for data data
results results QUAN qual
collection analysis follow-up collection analysis
Page 73
(c) Explanatory Design: Participant Selection Model
quan quan QUAL QUAL QUAL

quan QUAL Interpretation
data data participant data data
results results quan QUAL
collection analysis selection collection analysis
Figure 4.3 The Explanatory Design
73
persistence. In the second phase, she used a qualitative multiple case study
approach to help explain why certain factors identified in the first phase were
significant predictors of student persistence in the program.
The participant selection model (Figure 4.3c) is used when a researcher
needs quantitative information to identify and purposefully select participants
for a follow-up, in-depth, qualitative study. In this model, the emphasis of the
study is usually on the second, qualitative phase. For example, May and Etkina
(2002) collected quantitative data to identify physics students with consis-
tently high and low conceptual learning gains. They then completed an in-
depth qualitative comparison study of these students perceptions of learning.
Strengths of the Explanatory Design. The Explanatory Design is considered

the most straightforward of the mixed methods designs. The advantages of
this design include the following:
Its two-phase structure makes it straightforward to implement, because

the researcher conducts the two methods in separate phases and col-
lects only one type of data at a time. This means that single
researchers can conduct this design; a research team is not required
to carry out the design.
The final report can be written in two phases, making it straightfor-
ward to write and providing a clear delineation for readers.
This design lends itself to multiphase investigations, as well as single
mixed methods studies.
This design appeals to quantitative researchers, because it often begins
with a strong quantitative orientation.
Challenges in Using the Explanatory Design. Although the Explanatory

Design is straightforward, researchers choosing this approach still face chal-
lenges specific to this design.
For all variants:
This design requires a lengthy amount of time for implementing the

two phases. Researchers should recognize that the qualitative phase
(depending on the emphasis) will take more time than the quantitative
phase, but that the qualitative phase can be limited to a few partici-
pants. Still, adequate time must be budgeted for the qualitative phase.
The researcher must decide whether to use the same individuals for
both phases, to use individuals from the same sample for both phases,
or to draw participants from the same population for the two phases.
Chapter 6 explores approaches to using individuals from the same

sample or population in this approach.
It can be difficult to secure internal review board approval for this design
because the researcher cannot specify how participants will be selected
for the second phase until the initial findings are obtained. Approaches
to addressing this issue by tentatively framing the second, qualitative
phase for the internal review board are discussed in Chapter 6.
For the follow-up explanations model:
The researcher must decide which quantitative results need to be fur-

ther explained. Although this cannot be determined precisely until
after the quantitative phase is complete, options, such as selecting sig-
nificant results and strong predictors, can be discussed and weighed
as the study is being planned, as discussed further in Chapter 6.
For the participant selection model:
Investigators need to specify criteria for the selection of participants

for the qualitative phase of the research. Options include the use of
demographic characteristics, groups used in comparisons during the
quantitative phase, and individuals who vary on select predictors.
The Exploratory Design
As with the Explanatory Design, the intent of the two-phase Exploratory

Design (see Figure 4.4a) is that the results of the first method (qualitative)
can help develop or inform the second method (quantitative) (Greene et al.,
1989). This design is based on the premise that an exploration is needed for
one of several reasons: Measures or instruments are not available, the vari-
ables are unknown, or there is no guiding framework or theory. Because this
design begins qualitatively, it is best suited for exploring a phenomenon
(Creswell, Plano Clark, et al., 2003). This design is particularly useful when a
researcher needs to develop and test an instrument because one is not avail-
able (Creswell, 1999; Creswell et al., 2004) or identify important variables to
study quantitatively when the variables are unknown. It is also appropriate
when a researcher wants to generalize results to different groups (Morse,
1991), to test aspects of an emergent theory or classification (Morgan, 1998),
or to explore a phenomenon in depth and then measure its prevalence.
76
(a) Exploratory Design
04-Creswell (Designing)-45025.qxd
QUAL quan QUAL quan

results
5/16/2006
(b) Exploratory Design: Instrument Development Model
qual qual QUAN QUAN

8:35 PM
qual Develop QUAN Interpretation

data data data data
results instrument results qual QUAN
collection analysis collection analysis
Page 76
(c) Exploratory Design: Taxonomy Development Model
Develop
QUAL QUAL quan quan
QUAL taxonomy quan Interpretation
data data data data
results or theory results QUAL qual
collection analysis collection analysis
for testing
Figure 4.4 The Exploratory Design

Exploratory Design Procedures. Like the Explanatory Design, the

Exploratory Design is also a two-phase approach, and writers refer to it as
the Exploratory Sequential Design (Creswell, Plano Clark, et al., 2003). This
design starts with qualitative data, to explore a phenomenon, and then builds
to a second, quantitative phase (see Figure 4.4a). Researchers using this
design build on the results of the qualitative phase by developing an instru-
ment, identifying variables, or stating propositions for testing based on an
emergent theory or framework. These developments connect the initial qual-
itative phase to the subsequent quantitative component of the study. Because
the design begins qualitatively, a greater emphasis is often placed on the qual-
itative data. Myers and Oetzels (2003) study in appendix D on organizational
assimilation is an example of an Exploratory Design. They first explore the
topic qualitatively and develop themes from their qualitative data. They then
develop an instrument based on these results and subsequently use this
instrument in the second, quantitative phase of the study.
Variants of the Exploratory Design. This design has two common variants:
the instrument development model and the taxonomy development model.
Each of these models begins with an initial qualitative phase and ends with a
quantitative phase. They differ in the way the researcher connects the two
phases (see center boxes of Figures 4.4b and 4.4c) and in the relative empha-
sis of the two methods.
Researchers use the instrument development model (see Figure 4.4b)
when they need to develop and implement a quantitative instrument based on
qualitative findings. In this design, the researcher first qualitatively explores the
research topic with a few participants. The qualitative findings then guide
the development of items and scales for a quantitative survey instrument.
In the second data collection phase, the researcher implements and validates
this instrument quantitatively. In this design, the qualitative and quantitative
methods are connected through the development of the instrument items.
Researchers using this variant often emphasize the quantitative aspect of the
study. Using this model, Mak and Marshall (2004) initially qualitatively explored
young adults perceptions about the significance of the self to others in roman-
tic relationships (that is, how they perceive that they matter to someone else).
Based on their qualitative results, they developed an instrument and then
implemented it during a second quantitative phase in their study.
The taxonomy development model (see Figure 4.4c) occurs when the ini-
tial qualitative phase is conducted to identify important variables, develop a
taxonomy or classification system, or develop an emergent theory, and the sec-
ondary, quantitative phase tests or studies these results in more detail (Morgan,
1998; Tashakkori & Teddlie, 1998). In this model, the initial qualitative phase
produces specific categories or relationships. These categories or relationships

are then used to direct the research questions and data collection used in the
second, quantitative phase. This model is used when a researcher formulates
quantitative research questions or hypotheses based on qualitative findings
and proceeds to conduct a quantitative study to answer the questions. In addi-
tion, a researcher may identify emergent categories from the qualitative data
and then use the quantitative phase to examine the prevalence of these cate-
gories within different samples (Morse, 1991) or use taxonomy affiliation as a
basis for identifying comparison groups. For example, Goldenberg, Gallimore,
and Reese (2005) describe how they identified new variables and hypotheses
about predictors of family literacy practices based on their qualitative case
study. They then conducted a quantitative path analysis study to test these
qualitatively identified variables and relationships.
Strengths of the Exploratory Design. Due to its two-phase structure and the
fact that only one type of data is collected at a time, the Exploratory Design
shares many of the same advantages as the Explanatory Design. Its advan-
tages include the following:
The separate phases make this design straightforward to describe,

implement, and report.
Although this design typically emphasizes the qualitative aspect,
the inclusion of a quantitative component can make the qualitative
approach more acceptable to quantitative-biased audiences.
This design is easily applied to multiphase research studies in addi-
tion to single studies.
Challenges in Using the Exploratory Design. There are a number of chal-

lenges associated with the Exploratory Design and its variants.
For all variants:
The two-phase approach requires considerable time to implement.

Researchers need to recognize this factor and build time into their
studys plan.
It is difficult to specify the procedures of the quantitative phase when
applying for initial internal review board approval for the study.
Providing some tentative direction in a project plan for the internal
review board will be discussed further in Chapter 6.
Researchers should discuss whether the same individuals will serve as
participants in both the qualitative and quantitative phases (see the
use of different participants that we propose in Chapter 6).
For the instrument development model:
The researcher needs to decide which data to use from the qualitative
phase to build the quantitative instrument and how to use these data
to generate quantitative measures. In Chapter 7, we will discuss pro-
cedures for using qualitative quotes, codes, and themes to generate
aspects of quantitative instruments.
Procedures should be undertaken to ensure that the scores developed
on the instrument are valid and reliable. In Chapter 6, we will review
rigorous steps of instrument and scale development for this process.
For the taxonomy development model:
Decisions must be made in determining the relevant qualitative

findings to use. Options include using themes for variables and the
relationships between themes and subthemes (codes) for taxonomy
development.
SELECTING A TYPE OF MIXED METHODS DESIGN

Rigorous, high-quality studies result from well-designed research proce-
dures. Mixed methods researchers should select a specific design to use in
their studies. We often find researchers wanting to use more than one of the
four major designs in a study or to blend different aspects of the designs
together. However, we strongly recommend that researchers carefully select
a single design that best matches the research problem. This will make the
study more manageable and simpler to implement and describe. In addition,
it provides the researcher with a framework and logic to guide the imple-
mentation of the research methods.
What are the key factors that researchers should consider when choos-
ing a mixed methods design for their studies? Researchers should consider
the research problem that they want to study. A primary consideration is that
the design should match the research problem, as discussed in Chapter 2. In
addition, researchers should evaluate their own expertise and consider the
quantitative and qualitative skills that they possess. If they lack expertise with
certain methods (e.g., quantitative survey skills or collection of qualitative
field notes), they should consider working in a team or selecting a design
that does not emphasize that method. Consideration must also be given to
the available resources, such as the length of time available to complete the
study and funding resources for work in a team or the hiring of research
(a) What will the timing of the quantitative and qualitative methods be?
Concurrent Sequential
timing timing
Quantitative Qualitative
first first
(b) What will the weighting of the quantitative and qualitative methods be?
Equal Unequal
weight weight
Quantitative Qualitative
emphasis emphasis
(c) How will the quantitative and qualitative methods be mixed?
Merge Embed Connect

the data the data the data
Merging Merging Embed Embed Quantitative Qualitative

results data qualitative quantitative leads to builds to
during during data in a data in a qualitative quantitative
interpretation analysis quantitative qualitative
design design
Figure 4.5 Decision Tree for Mixed Methods Design Criteria for Timing, Weighting, and Mixing
SOURCE: Based on Creswell, Plano Clark, et al. (2003); Hanson et al. (2005); and Plano Clark (2005).
assistants. The expectations of audiences for the research can also influence
the design choice, particularly if the audience values one type of evidence
over the other type.
In addition to these factors, the choice of a research design relates to
three decisions: the timing of the use of collected data (i.e., the order in
which the data are used in a study), the relative weight of the quantitative and
qualitative approaches (i.e., the emphasis given to each), and the approach
to mixing the two datasets (i.e., how the two datasets will be related or con-
nected). A decision tree, shown in Figure 4.5, can help identify choices for
each of these three decisions.
The Timing Decision
When selecting a mixed methods approach, researchers must answer the

question: What will the timing of the quantitative and qualitative methods be?
(Figure 4.5a). Timing (also referred to as implementation or sequence)
refers to the temporal relationship between the quantitative and qualitative
components within a study (Greene et al., 1989). Timing is often discussed
in relation to the time the data sets are collected. However, most importantly,
it describes the order in which the researchers use the data within a study
(Morgan, 1998). Therefore, timing relates more to when the data are ana-
lyzed and interpreted than to when the data are collected, although these
times are often interrelated.
As shown in Figure 4.5a, timing within a mixed methods design is classi-
fied in one of two ways: concurrent or sequential (Morse, 1991). Concurrent
timing occurs when the researcher implements both quantitative and qualita-
tive methods during a single phase of the research study. This means that the
quantitative and qualitative data are collected, analyzed, and interpreted at
(approximately) the same time. Sequential timing occurs when the researcher
implements the methods in two distinct phases, using (collecting and analyz-
ing) one type of data before using the other data type. There are two options
for sequential timing. A researcher may choose to start by collecting and ana-
lyzing quantitative data and may then subsequently collect and analyze quali-
tative data. The reverse is also possible: Qualitative data are collected and
analyzed first and then quantitative data are collected and analyzed.
The Weighting Decision
In addition to choosing the timing, researchers also need to consider

the relative weighting of the two approaches in the study (Figure 4.5b).
Weighting refers to the relative importance or priority of the quantitative and
qualitative methods to answering the studys questions. This choice has been
referred to as the priority decision (Morgan, 1998) because a researcher
decides whether both methods will have equal priority or one method will
have a greater priority than the other.
There are two possible weighting options for a mixed methods design, as
depicted in Figure 4.5b. The two methods may be given equal weight so that
both play an equally important role in addressing the research problem. On

the other hand, the research design may weight them unequally. In this case,
one of the methods (quantitative or qualitative) will have a greater emphasis
within the study than the other method (qualitative or quantitative).
How does a researcher select a studys weighting? Numerous consid-
erations influence the comparative weights of the qualitative and quantita-
tive data in a study. Morse (1991) suggested that the theoretical drive, or
worldview, used to guide a study determines its weighting. That is, a post-
positivistic worldview calls for a quantitative priority, a naturalistic world-
view calls for a qualitative priority, and a pragmatic worldview calls for
either equal or unequal weighting, depending on the research question.
Morgan (1998) advised that the weighting in a study be based on the
strength of which data collection method (quantitative or qualitative) is
best suited to address the studys goals or purpose. The weighting is thus
influenced by the goals, the research question(s), and the use of proce-
dures from research traditions such as quantitative experimental designs or
qualitative case study designs.
Practical considerations also influence weighting (e.g., Creswell, 2003).
For example, it takes more resources to implement a study that gives equal
weighting to the two methods. Therefore, with limited resources, a
researcher may choose to prioritize one method (quantitative or qualitative)
and devote fewer resources to the secondary method (qualitative or quanti-
tative). The weighting may also reflect the researchers relative experience
with the two methods, particularly if he or she is significantly more familiar
with one method than the other. Finally, consider the audiences for the
research. Audiences include advisors, committee members, journal editors
and reviewers, funding officers, and the disciplinary research community at
large. If a studys target audiences are unaccustomed to or unaccepting of
one approach (quantitative or qualitative), then the other method may
receive a greater priority in the studys design.
Researchers should indicate a studys weighting within their written
reports, and research consumers can look for these indications as they read
published mixed methods studies. Indicators of a studys weighting include
the following:
The way the researcher words the title: Quantitative or qualitative

terms indicate unequal weighting, and the lack of such terms indi-
cates equal weighting
The explicit identification of the guiding worldview used in the study
A purpose statement that uses terms that indicate unequal weighting,
such as primary aim or secondary purpose
A statement identifying the weighting in the methods section
More space being devoted to one method in the article or the empha-
sis of one method within the abstract
More sophisticated and complex procedures used for one method
compared to the other
The Mixing Decision
The third procedural consideration for choosing a mixed methods

design is how the quantitative and qualitative methods will be mixed (Figure
4.5c). Mixing is the explicit relating of the two data sets. A study that includes
both quantitative and qualitative methods without explicitly mixing the data
derived from each is simply a collection of multiple methods. A rigorous and
strong mixed methods design addresses the decision of how to mix the data,
in addition to timing and weighting.
What procedures are available for mixing quantitative and qualitative
data? Conceptually, there are three overall strategies for mixing quantitative
and qualitative data (see Figure 4.5c). The two data types can be merged, one
can be embedded within the other, or they can be connected.
Merging Data Sets. The data are merged when the researcher takes the two
data sets and explicitly brings them together or integrates them. Researchers
can merge the two data sets during the interpretation (by analyzing them
separately in a results section and then merging the two sets of results
together during the interpretation or discussion phase) or during the analy-
sis of the data (by transforming one data type into the other type or consoli-
dating the data into new variables).
Embedding Data at the Design Level. The researcher could decide to embed
data of one type within a design of the other type. This is an example of mix-
ing at the design level, not just at the level of data. A researcher may choose
to embed qualitative data within a larger quantitative (e.g., experimental)
design or to embed quantitative data within a larger qualitative (e.g., phe-
nomenology) design. One form of data can be embedded in a concurrent
data collection with the other dataset; alternatively, the embedded data may
be collected sequentially before or after the other dataset. Researchers may
make interpretations from using the secondary, embedded dataset by bring-
ing the two datasets together in the concurrent approach and keeping them
separate in the sequential approach.
Connecting From Data Analysis to Data Collection. A researcher could

choose to connect the two data types. Connecting the data occurs when the
analysis of one type of data leads to (and thereby connects to) the need for
the other type of data. This can occur in one of two ways. A researcher may
obtain quantitative results that lead to the subsequent collection and analysis
of qualitative data. A researcher can also start with qualitative results that
build to the subsequent collection and analysis of quantitative data. The mix-
ing occurs in the way that the two data types are connected. This connection
can occur in different ways, such as in specifying research questions, select-
ing participants, or developing an instrument or other materials.
IMPLEMENTING THE DESIGN DECISIONS

Researchers could choose to use any combination of timing, weighting, and
mixing for their mixed methods design. However, based on the underlying
logic of the mixed methods designs introduced in this chapter, these criteria
are best used in certain combinations. Table 4.2 summarizes the four major
designs and their corresponding timing, weighting, and mixing decisions.
These decisions, combined with the different research purposes, lead to the
following design choices:
If there is a single phase, both types of data are given equal emphasis, the
two sets of results are converged during the interpretation, and the intent
is to draw valid conclusions about a research problem, then the choice of
design is the Triangulation Designconvergence model.
If there is a single phase, both types of data are given equal emphasis,
one type of data is transformed into the other type, and the intent is
to interrelate different data types about a research problem, then the
choice of design is the Triangulation Designdata transformation model.
If both types of data are collected at the same time from a survey and
the intent is to use qualitative information to validate the quantitative
results, then the choice of design is the Triangulation Designvalidat-
ing quantitative data model.
If different types of data are collected to represent different levels of
analysis within a system, with the intent of forming an overall inter-
pretation of the system, then the choice of design is the Triangulation
Designmultilevel model.
If quantitative data are used to answer the primary question in an exper-
imental design and qualitative data are embedded within the experi-
mental design (before, during, or after the intervention) with the intent
of answering a secondary question related to the experiment, then the
choice of design is the Embedded Designexperimental model.
Table 4.2 The Major Mixed Methods Design Types
Design Type Variants Timing Weighting Mixing Notation

Triangulation Convergence Concurrent: Usually Merge the QUAN +
Data quantitative equal data during the QUAL
transformation and interpretation
Validating qualitative at or analysis
quantitative same time
data
Multilevel
Embedded Embedded Concurrent or Unequal Embed one QUAN(qual)
experimental sequential type of data or
Embedded within a larger QUAL(quan)
correlational design using
the other type
of data
Explanatory Follow-up Sequential: Usually Connect the QUAN qual
explanations Quantitative quantitative data between
Participant followed by the two
selection qualitative phases
Exploratory Instrument Sequential: Usually Connect the QUAL quan
development Qualitative qualitative data between
Taxonomy followed by the two
development quantitative phases
If quantitative data are used to answer the primary question in a

correlational design and qualitative data are embedded within the cor-
relational design with the intent of explaining the mechanisms that
relate the predictor and outcome variables, then the choice of design
is the Embedded Designcorrelational model.
If one phase is followed by another phase, the first phase is quantita-
tive, quantitative methods or data are emphasized, the second phase
is connected to the results of the first phase, and the intent is to
explain these results using qualitative data as a follow-up, then the
choice of design is the Explanatory Designfollow-up explanations
model.
If one phase is followed by another phase, the first phase is quantitative,
the qualitative phase is emphasized, the second phase is connected to
the results of the first phase, and the intent is to purposefully select
participants to best address the qualitative research question, then
the choice of design is the Explanatory Designparticipant selection
model.
If one phase is followed by another phase, the first phase is qualita-
tive, the two phases are connected by the development of an instru-
ment based on the results of the first phase, and the intent is to
develop and implement an instrument on the topic of interest, then
the choice of design is the Exploratory Designinstrument develop-
ment model.
If one phase is followed by another phase, the first phase is qualitative
and results in a taxonomy or emergent theory, the two phases are con-
nected by quantitative testing of the results of the first qualitative
phase, the qualitative phase is emphasized, and the intent is to quan-
titatively generalize the qualitative results, then the choice of design is
the Exploratory Designtaxonomy development model.
WRITING A PARAGRAPH TO
IDENTIFY A STUDYS DESIGN
Because many researchers and reviewers are currently unfamiliar with the
different types of mixed methods designs, it is important to include an
overview paragraph that introduces the design when writing about a study in
proposals or research reports. This overview paragraph generally is placed at
the start of the methods discussion and should address four topics. First,
identify the type of mixed methods design and variant model, if appropriate.
Next, give the defining characteristics of this design, including its timing,
weighting, and mixing decisions. Third, state the overall purpose or rationale
for using this design for the study. Finally, include references to the mixed
methods literature on this design. An example of an overview paragraph is
included in Figure 4.6, along with comments that will assist in identifying
these features within the paragraph.
Summary
Researchers designing a mixed methods study can choose among the

four major types of mixed methods designs: Triangulation, Embedded,
Names design
Mixed Methods Sequential Explanatory Design
The mixed methods sequential explanatory design consists of two distinct

phases: quantitative followed by qualitative (Creswell, Plano Clark, et al., Discusses timing, weighting
2003). In this design, a researcher first collects and analyzes the quantita-
tive (numeric) data. The qualitative (text) data are collected and analyzed
second in the sequence and help explain, or elaborate on, the quantitative Discusses mixing
results obtained in the first phase. The second, qualitative, phase builds on
the first, quantitative, phase, and the two phases are connected in the inter-
mediate stage in the study. The rationale for this approach is that the quan-
Discusses design rationale
titative data and their subsequent analysis provide a general understanding
of the research problem. The qualitative data and their analysis refine and
explain those statistical results by exploring participants views in more depth
(Rossman & Wilson, 1985; Tashakkori & Teddlie, 1998; Creswell, 2003). Cites methodological
references
Figure 4.6 Sample Paragraph Introducing a Mixed Methods Design

SOURCE: Ivankova et al. (2006, p. 5).
Explanatory, or Exploratory. Mixed methods researchers choose a design

based on which design best addresses the research problem and the advan-
tages inherent in each design. Researchers should carefully consider the chal-
lenges associated with their design choice and plan strategies for addressing
these challenges. As part of choosing a design, decisions need to be made
about the use of concurrent or sequential timing for the two methods,
whether the two methods will have equal or unequal weighting, and how the
two methods will be mixed. These decisions, the underlying logic that is best
suited to the research problem, and practical considerations are the founda-
tion researchers should use in selecting which variant of the four major
mixed methods designs to use for their study.
Activities
1. Which of the four major design types will you use in your study? Write
a one-paragraph overview that identifies this design; defines its
timing, weighting, and mixing; and states your rationale for choosing
it for your study.
2. What challenges are associated with your design choice? Write a para-
graph that discusses the challenges that you anticipate occurring with
your design and how you might address them.
3. Draw a diagram of the procedures you will use, following one of the
major variants of the four types of designs advanced in this chapter.
Use the depictions of the different variants in the figures as examples
to follow.
Additional Resources to Examine
For additional information on the major mixed methods design types, consult
Creswell, J. W., Plano Clark, V. L., Gutmann, M., & Hanson, W. (2003). Advanced mixed
methods research designs. In A. Tashakkori & C. Teddlie (Eds.), Handbook of
mixed methods in social and behavioral research (pp. 209240). Thousand
Oaks, CA: Sage.
Greene, J. C., Caracelli, V. J., & Graham, W. F. (1989). Toward a conceptual framework
for mixed-method evaluation designs. Educational Evaluation and Policy
Analysis, 11(3), 255274.
Good discussions on timing, weighting, and mixing in mixed methods stud-

ies can be found in
Caracelli, V. J., & Greene, J. C. (1993). Data analysis strategies for mixed-method eval-
uation designs. Educational Evaluation and Policy Analysis, 15(2), 195207.
Morgan, D. L. (1998). Practical strategies for combining qualitative and quantitative
methods: Applications to health research. Qualitative Health Research, 8(3),
362376.
Morse, J. M. (1991). Approaches to qualitative-quantitative methodological triangula-
tion. Nursing Research, 40, 120123.
Tashakkori, A., & Teddlie, C. (1998). Mixed methodology: Combining qualitative and
quantitative approaches. Thousand Oaks, CA: Sage.
CHAPTER 5
INTRODUCING A MIXED
METHODS STUDY
A fter we understand the characteristics of mixed methods research,

assess the preliminary considerations, review studies, and select a
research design, we may begin the more detailed process of design-
ing and conducting a mixed methods study. This chapter begins with the first
steps in this process: developing an introduction to a mixed methods study.
There are several parts to this introduction: writing a title, developing a state-
ment of the problem section, composing a purpose statement, and detailing
research questions or hypotheses. These introductory parts begin the write-
up of a study, forming the opening paragraphs of a journal article or the first
chapter in a dissertation or thesis. For each of these parts, we will highlight
two strategies in this chapter: adapting it to mixed methods research and
relating it to the design chosen in Chapter 4. To accomplish this, we will use
appropriate mixed methods terms and design some elements of research
that may be new to the reader, such as a purpose statement, which will be
scripted with mixed methods components for a type of design; research
questions, which will be written from the perspective of the mixed methods
design; and a statement of reasons for the selection of a particular mixed
methods design. Writing about mixed methods and the specific design is
important because it foreshadows the methods and provides a means by
which readers can evaluate the study.
89

MIxed Method Design PDF

Uploaded by

Copyright:

Available Formats

MIxed Method Design PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

MIxed Method Design PDF

Uploaded by

Copyright:

Available Formats

04-Creswell (Designing)-45025.

qxd 5/16/2006 8:35 PM Page 58

R esearch designs are procedures for collecting, analyzing, interpreting,

The classifications of designs in the literature

Choosing a Mixed Methods Design 59

Factors such as timing, weighting, and mixing, which influence the

CLASSIFICATIONS OF MIXED METHODS DESIGNS

THE FOUR MAJOR TYPES OF MIXED METHODS DESIGNS

60 DESIGNING AND CONDUCTING MIXED METHODS RESEARCH

Table 4.1 Mixed Method Design Classifications

Author Mixed Method Designs Discipline

Greene, Caracelli, Initiation Evaluation

Patton (1990) Experimental design, qualitative data, and Evaluation

Morse (1991) Simultaneous triangulation Nursing

Steckler, McLeroy, Model 1: Qualitative methods to develop Public health education

Greene and Component designs Evaluation

Choosing a Mixed Methods Design 61

Author Mixed Method Designs Discipline

62 DESIGNING AND CONDUCTING MIXED METHODS RESEARCH

Table 4.1 (Continued)

Author Mixed Method Designs Discipline

The Triangulation Design

The most common and well-known approach to mixing methods is the

Triangulation Design Procedures. The Triangulation Design is a one-phase

Choosing a Mixed Methods Design 63

(a) Triangulation Design

(b) Triangulation Design: Convergence Model

QUAL QUAL Transform

(d) Triangulation Design: Validating Quantitative Data Model

QUAN data QUAN

Figure 4.1 (Continued)

64 DESIGNING AND CONDUCTING MIXED METHODS RESEARCH

(e) Triangulation Design: Multilevel Model

Figure 4.1 The Triangulation Design

Choosing a Mixed Methods Design 65

confirm, or corroborate quantitative results with qualitative findings. The

66 DESIGNING AND CONDUCTING MIXED METHODS RESEARCH

Strengths of the Triangulation Design. This design has a number of strengths

The design makes intuitive sense. Researchers new to mixed methods

Challenges in Using the Triangulation Design. Although this design is the

Much effort and expertise is required, particularly because of the con-

For the convergence model:

Researchers need to consider the consequences of having different

Choosing a Mixed Methods Design 67

and qualitative data are usually collected for different purposes

For the data transformation model:

Researchers need to develop procedures for transforming data and

The Embedded Design

Embedded Design Procedures. The Embedded Design mixes the different

68 DESIGNING AND CONDUCTING MIXED METHODS RESEARCH

(a) Embedded Design

QUAN Interpretation QUAL Interpretation

(b) Embedded Design: Embedded Experimental Model

(c) Embedded Design: Embedded Correlational Model

Figure 4.2 The Embedded Design

in a phenomenology design (see Figure 4.2a). The Embedded Design

Choosing a Mixed Methods Design 69

Variants of the Embedded Design. Although many variants of the Embedded

70 DESIGNING AND CONDUCTING MIXED METHODS RESEARCH