Mixed methods: Integration of qualitative & quantitative data collection and analysis in a single inquiry Arden M. Morris, MD, MPH Associate Professor of Surgery Center for Health Outcomes and Policy University of Michigan
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Mixed methods: Integration of qualitative &
quantitative data collection and analysis in a single inquiry
Arden M. Morris, MD, MPH Associate Professor of Surgery
Center for Health Outcomes and Policy University of Michigan
No relevant financial disclosures
Partly supported by the Michigan Institute for Clinical & Health Research Summer
Disparities Immersion Program
Associated with the UW Collaborative to Improve Native Cancer Outcomes
(CINCO)
Raven: A Trickster Tale from The Pacific Northwest by Gerald McDermott
Objectives
• Describe MMR: rationale, design, benefits and challenges
• Review concepts using a systematic literature review example
• Discuss how to present and evaluate MMR studies
What is MMR?
Quantitative • Small number variables, large datasets • Closed ended Qs • Statistical analyses of numbers: Point
estimate for the mean of the population • Emphasizes generalizability • Concerns for bias
Qualitative • Small sample size • Open ended Qs • Inductive or deductive analysis of words • Emphasizes explanation (meaning,
model, mechanism, relationships, context) • Concerns for bias
MMR: What why when
• Data must be integrated, compared, contrasted, appraised, & synthesized
• Quantitative & qualitative approaches are complementary
• Neither quantitative nor qualitative approach is sufficient in capturing all relevant information
Six core characteristics of MMR
• Collection of qualitative and quantitative data (open- and closed-ended) in response to research Qs
• Analysis of both qualitative and quantitative data • Persuasive and rigorous procedures for the qualitative
and quantitative methods • The integration of these two data sources (merging,
connecting, embedding) • The use of a specific MM design using concurrent or
sequential integration • A philosophical foundation
9
Creswell, 2010
10
Post-positivism Determination Reductionism Empirical observation and measurement Theory verification
Constructivism Understanding Multiple participant meanings Social and historical construction Theory generation
Participatory Political Empowerment issue-oriented Collaborative Change-oriented
Pragmatism Consequences of actions Problem-centered Pluralistic Real-world practice oriented
Creswell, 2010
Mixing the methods Timing, implementation, priority
11
Connect data
Converge data
Embed the data
Quan Qual
Results Qual Quan
Results Qual Quan
Mixed methods designs
• Convergent parallel design • Explanatory sequential design • Exploratory sequential design • Embedded design • Transformative design • Multiphase design
Creswell, 2010
13
Convergent Parallel Design
Embedded Design
Concurrent MMR Designs
Creswell, 2010
QUAN Data collection
& analysis
Interpretation QUAL
Data collection & analysis
Interpretation
Quan Pre-test data & analysis Qual
process
Quan Post-test data
& analysis
intervention
14
Explanatory Design
Exploratory Design
Sequential MMR Designs - 1
Creswell, 2010
QUAN Data collection
& analysis
Interpretation qual
Data collection & analysis
QUAL Data collection
& analysis
Interpretation quan
Data collection & analysis
15
Sequential Embedded Design
Sequential MMR Designs - 2
Creswell, 2010
Qual Pre-
intervention
Interpretation QUAN
Intervention trial
Qual Post-
intervention
16
Sequential Embedded Design
Transformative Design
Qual Pre-
intervention
Interpretation QUAN
Intervention trial
Qual Post-
intervention
Multiphase (or Multi-project) Design
17
Interpretation quan
Data collection & analysis
QUAL Data collection
& analysis
QUAN Data collection
& analysis
Interpretation QUAL
Data collection & analysis
Technique Developments
• Types of designs; diagrams, detailed procedures, notation
• Scripts for purpose statements • Mixed methods research questions • Analysis strategies for merging data • Point of interface strategies for
sequential data
18
Typical Scenarios
• Survey & focus groups data merged and compared • Survey completed then focus groups used to
explain the quantitative results • Focus groups conducted and data used to construct
follow-up survey with a large sample • Qualitative data are collected before and after an
experiment is conducted • A longitudinal study with multiple quantitative and
qualitative studies to address a single overarching research objective
Example project: Systematic literature review
and meta-synthesis
Evidence of a problem • Planning CINCO MMR project
– Focus groups (tribe members) – survey/interviews (MDs)
• What is known about shared decision making among AI/AN patients?
• How should we incorporate the domain of SDM into survey?
Shared decision making
Patient choice
Shared decision making
Provider choice
Starting with a question
How does SDM happen among AI/AN ca pts?
What is the process & who makes
decisions among AI/AN ca pts?
Starting with a question
What is the process & who makes
decisions among AI/AN ca pts?
What is the process & who makes
decisions among minority ca pts?
Developing the taxonomy
• SDM: 2-way information exchange btw MD & pt followed by discussion of tx preferences until they reach consensus on a tx decision
• Racial/ethnic or cultural minority group: A racial, ethnic, or cultural minority population in the country where the study took place.
Inclusion Criteria Exclusion Criteria
Fit the definition of SDM DM for cancer treatment in children
Reported data on cancer treatment DM
Data not stratified by race/ethnicity or no minority group
Collected primary data Palliative or end-of-life care in cancer treatment
Results related to DM in a minority group
Peer-reviewed publication
Data acquisition
Data abstraction • A data abstraction tool developed and
iteratively revised • Across-method and method-specific study
features, sampling strategy, methods, results, conclusions, and assessments of methodological quality
• 2 reviewers independently data points and results were combined
Data abstraction
Data abstraction - 2
Data synthesis • Iterative thematic analysis • Conceptual model development • Findings synthesized for all cancer sites, sorted
and summarized by theme and study type (quan vs. qual).
• Confidence intervals were reported if available • Racial, ethnic, and gender differences described
Thematic analysis 1. Familiarization w data: 2 reviewers independently
read each paper in-depth 2. Data were independently coded 3. Reviewers searched the data for significant
themes 4. Reviewers discussed, compared and contrasted
the themes across studies and revised until 5. Consensus reached on final data-driven themes 6. Organized into a conceptual model
Level Theme - 1 Quan Qual
DM Process
Decisional role 8 11
Conflict btw actual & pref decisional role 2 5
Satisfaction/regret w DM or decision 3 6
Patient Factors
Spirituality and fatalism 1 6
Attitudes about tx & decision 3 3
Self-efficacy 2 1
Level of acculturation & language 6 5
Level Theme - 2 Quan Qual
Family/ Impt Others Factors
Family & others’ participation or advocacy & DM 6 8
Influence of family, friends, & others’ experiences 1 3
Participation or advocacy of others and receipt of treatment 6 0
Consequences of ceding decision-making autonomy 0 8
Community Factors
DM as a collective experience 0 3
Cultural/community norms & values 0 6
Communication w the community 0 4
Level Theme - 3 Quan Qual
Provider Factors
Provider preferences and recommendations
4 5
Provider communication & information-giving
3 8
Conflict and cultural congruence in the patient-provider relationship
0 3
A revised conceptual model
Linkages between themes
• Social support • Communication • Cultural congruence
Reporting and Assessing MMR
Challenges to reporting MMR
• Audiences familiar with only one approach • Different language and terminology • Journal word counts • Reviewer inexperience and bias
6 criteria • Identify content or context • Rationale for mixing • Describe data types & sampling design • Priority of qualitative & quantitative • Implementation sequence • Design model
Wisdom et al 2012 Creswell et al 2004
Reporting MMR
• Hard to satisfy both experts & novices • Educate reviewers—consider providing
similar published studies as appendices • Trial run figures/tables among a few people
you trust • Onus is on you to communicate clearly &
within bounds
Final thoughts • Clarify your research question • Access expertise in both methods • Take the time to create and test conceptual
models • Understand the taxonomy and consider a
pragmatic approach • Refer to successfully published MMR studies
Thanks!
Related Documents
