Professional Documents
Culture Documents
Defining mixed methods research is not an easy task because there are a multitude
of design variations, paradigmatic combinations, and sources of quantitative or qualitative
data. In the research field, there is continuing discussion regarding clarification of mixed
methods research in relationship to mono method research, multi method research, and
mixed model research.
The concept of purpose gets to the heart of why a mixed methods approach is chosen:
whether to seek corroboration of findings, to elaborate or clarify findings, to further
develop interpretations, to investigate contradictions, or to expand the breadth or depth of
a study. The five general purposes discussed here were identified by Greene, Caracelli,
and Graham (1989) and have been reiterated by others.
Development uses the results from one approach to develop or inform the other
approach.The study can begin with qualitative data analyzed in the first phase as a way to
establish constructs to be measured quantitatively in the second phase. Alternatively,
quantitative survey results from an initial phase of a study might be used for purposeful
selection of participants for a second qualitative phase.
Expansion is focused on expanding the breadth and range of the inquiry by using
different methods for different components of the study. For example, a researcher might
be interested in examining the effectiveness of a professional development program in
raising student achievement. Qualitative observations might be used to determine whether
and how well the teachers are using the professional development approaches in the
classroom, which is sometimes referred to as implementation fidelity. Quantitative test
scores could be used as a quantitative measure of program effectiveness.
1. The number and type of data collection approaches that will be used
How the researcher answers these questions determines the most appropriate mixed
methods design to be used. Six mixed methods designs are described here.
Parallel designs are those in which data are collected and analyzed separately,
similar to concurrent designs. However, in concurrent designs, inferences are made in a
more integrated manner, whereas in parallel designs each data set leads to its own set of
inferences. Indeed, sometimes the results may be reported in two separate write-ups in the
same report. Some authors have referred to this design as a quasi-mixed method design.
In sequential designs, data that are collected and examined in one stage inform the
data collected in the next phase. Data analysis begins before all data are collected and the
analysis may influence choices made in conducting the next phase of the study. Each
separate phase may shape the conceptual and methodological approaches used in
following phases. Different forms of data are collected in sequence at different phases in a
study. For example, a quantitative survey collected and analyzed in the first phase may
inform the second qualitative phase of the study, or qualitative observations conducted in
the first phase may inform development of a quantitative survey in the second phase. The
sequential order- ing provides information necessary to conduct a more thorough study.
Fully mixed designs or fully integrated designs involve mixing of the qualitative
and quantitative approaches in an interactive way throughout the study. At each stage, one
approach may influence the implementation of the other. There are multiple points of
integration throughout the study, from data collection to data analysis and interpretation
and inferences.
Conversion designs involve transforming data, which means that data collected in
one form (e.g., numbers or text) are converted to a different form and then analyzed.
Qualitative data may be converted to quantitative data or vice versa. Data are analyzed
from both perspectives and inferences are made based on both sets of analyses. For
example, qualitative data might be quantitized by counting the number of times a
particular word is used or the number of times a particular theme is identified.
Quantitative data from a survey might be used to create a written profile of a group. Data
transformation is a term used in mixed methods research to describe when a researcher
begins with qualitative data, such as interviews or secondary source documents, and
through analysis trans- forms the words into numerical values for comparative or
statistical analysis. The transformed data are then compared with data from a comparable
method (quantified qualitative data compared with statistical results from a separate
phase, or qualitized descriptive statistics from a survey compared to coded themes that
emerge from analysis of interviews). Conversion designs require that at least one of the
data sources be transformed into another type of data. Some researchers argue that it is
inappropriate to quantify qualitative data because it considerably reduces the value of the
data as a tool for understanding a phenomenon.
In an embedded design, one form of data supports a second form of data within a
single study. For example, if the purpose for the research project is to inform
administration or change policy, quantitative and qualitative data may be required to
convince those with the power to make changes that the results are credible. The rationale
for an embedded design is that a single data set is not sufficient to answer different
questions, and each type of question requires different types of data. The most common
form of embedded mixed methods research is when quantitative data are embedded
within a qualitative case study and the quantitative data are supportive of the major
qualitative findings. Alternatively, qualitative data can be supportive of statistical results
by addressing questions that are unanswerable using experimental or correlation research.
The hallmark of an embedded design is that the different data sets are mixed at the design
level rather than in the conclusions so that one of the data sources is relegated to playing a
supplemental role to the prominent data source. One challenge that must be dealt with
when conducting embedded research is that the purpose of the study may not be clear to
the participants or research team members involved in the project.
NOTATION SYSTEM
The most widely accepted notation system used in mixed methods designs, first
introduced by Morse (1991), uses a plus sign (+) to indicate that the data collection and
analysis of methods occur at the same time. An arrow (→) indicates that data collection
and analysis occur in sequence. Morse also indicates that the weight or importance of the
methods within the study should be denoted by using uppercase letters for prominence
and lowercase letters to indicate less dominant methods. There are no specific rules that
determine appropriate proportions of qualitative and quantitative research in a mixed
methods study. Some researchers use parentheses to indicate methods that are embedded
within other methods. The notation system rules are shown here
Weighting priority
QUAL+QUAN (both are equally important) QUAN+qual (quantitative approach is
dominant) QUAL+quan (qualitative approach is dominant)
Sequence
QUAN → Qual (quantitative collection or analysis occurs first followed by qualitative
collection/analysis)
QUAL → Quan (qualitative collection or analysis occurs first followed by quantitative
collection/analysis)
Visually representing a mixed methods design is a useful tool for helping the
researcher to determine and show logic in the relationship between the components. The
visual interactive model allows for a clear representation of how the purpose, conceptual
framework, methodology, methods, and issues including validity/credibility interact with
the central overarching research question.
Stage 1, data reduction, involves analyzing the qualitative data via theme analysis or
thematic coding while also analyzing the quantitative data via descriptive statistics, factor
analysis, etc.
Stage 2, data display, involves using tables and graphs to display the quantitative data
and using other forms, such as matrices, rubrics, and lists, to describe the qualita- tive
data.
Stage 4 is data correlation, which involves comparing the data from the different analy-
ses (quantitizing and qualitizing compared to the originals). The authors use the term
correlation in a different sense than typically considered in quantitative research. Here,
the researcher is comparing, for example, the original qualitative data and the quantitized
qualitative data to determine whether the two sets seem to reflect similar findings.
Stage 5 is data consolidation, in which both sets of data are combined to create a new set
of data or variables.
Stage 6, data comparison, involves comparing data from the qualitative and quantitative
data sources.
Finally, stage 7 is data integration, in which the data and interpretations are integrated
into either a coherent whole or reported in two separate sets (qualitative and quantitative)
of coherent wholes. Figure 19.1 and Table 19.2 present qualitative data (McKenzie, 2008)
that were quantitized in order to show how the data transformation stage might work.
ohnson, R. B., & Onwuegbuzie, A. J. (2004). Mixed methods research: A research paradigm
whose time has come. Educational Researcher, 33(7), 14–26.
Creswell, J. & Plano Clark, V. (2007). Designing and conducting mixed methods research.
Thousand Oaks, CA: Sage.