Mixed Methods Procedures in Research Design

By John W. Cresswell

Mixing quantitative and qualitative data in single study
- Inclusion of multiple method of data
- Multiple form of analysis
- Call for explicit procedure
- Focused on understanding a problem of using two methods
- Rationale for using multiple forms of data collection and analysis
Components of mixed methods procedures p. 209
1. Visual model
2. Specific procedures of data collection and analysis
3. Researcher’s role
4. Structure for presenting final report
Nature of mixed method research
- Evolution in psychology and in the multitrait-multimethod in converging or triangulation
- Reasons
(1) To expand understanding (from one method to another)
(2) To converge r confirm finding (from diff. data source)
- Challenges …Need for
(1) Extensive data collect
(2) Time intensive of analysis both text and numeric data
(3) Familiar with both quan and qual
Types of mixed methods strategies p. 210
- Criteria for choosing a strategy
1. What is the implementation sequence of quan. and qual. data collection
2. What priority will be given to quan. and qual. data collection and analysis
3. What stage in research project will quan. and qual. data and finding integrated
4. Theoretical perspective be used in the study
Implementation p. 211
- Collect data in phase (sequential) or gather at the same time (concurrently)
- จะเก็บข้อมูลเชิงคุณภาพ หรือเชิงปริมาณก่อน ขึ้นอยู่กับนักวิจัย
- Qual data –first: to explore the topic
Priority for one type depends on p. 212
- Interest of researcher
- Audience for the study
- What seek to emphasize in the study
- Extent of treatment of data
- Use of theory (Inductive or deductive)
Integration of two types of data
- Occur at several stages เช่น data collection, data analysis, interpretation, some combination of places
- Alternative strategies and visual models Example
(1) Data collection: combining open-ended and close-ended questions
(2) Data analysis / interpretation เช่น transform qual. themes or codes into quan.
A theoretical perspective
- Whether theoretical perspective guides the entire design
- Perspective may be one from social science or from advocacy/participatory lens (gender, race, class)
Proposal contain
1. Description of strategy
2. Visual model
3. Basic procedures (that use in implementing strategy
Six major strategies
1. Sequential Explanatory
2. Sequential Exploratory
3. Sequential Transformative
4. Concurrent Triangulation
5. Concurrent Nested
6. Concurrent Transformative
1. Sequential Explanatory p. 215
- Priority to quan
- เก็บ และวิเคราะห์ข้อมูลแบบ quan ก่อน แล้วจึงเก็บ/วิเคราะห์แบบ qual
- Integrated in interpretation
- Purpose: ใช้ คุณภาพ ช่วยในการอธิบายและตีความข้อค้นพบจากปริมาณ เช่น unexpected result หรือ อธิบาย more detail
- เหมาะกับการ explaining and interpreting relationships
- จุดอ่อน: ใช้เวลานาน (การเก็บข้อมูล) และการแยกเป็นสองช่วง (two separate phases)จะมีปัญหา ถ้าทั้งสองช่วงมีความสำคัญเท่าๆกัน
2. Sequential Exploratory
- Many features similar to the Sequential Explanatory
- Conducted in two phases
- Initial of qual. Follow by quan.
- Priority to qual.
- Integrated in interpretation
- Use quan. Data and results to assist in interpretation of qual. Findings
- Purpose:
(1) to explore phenomenon
(2) testing elements of emergent theory
(3) generalize qual finding to diff. samples
(4) develop and test instrument
- advantages
(1) expand on qual. Findings
(2) building a new instrument
(3) make qual. Study ให้เป็นที่ยอมรับมากขึ้น
- จุดอ่อน: ใช้เวลานาน (ในการเก็บข้อมูล) และค่อนข้างยากในการเปลี่ยนจากการวิเคราะห์เชิงคุณภาพไปสู่การเก็บข้อมูลเชิงปริมาณ
3. Sequential Transformative p. 216
- Two distinct data collection phases
- ใช้คุณภาพ / ปริมาณ นำก่อนก็ได้ หรือจะให้ความสำคัญเท่าๆกันก็ได้ ถ้ามีทรัพยากรเพียงพอ
- Integrated in interpretation
- Has theoretical perspective guide the study
- Purpose: employ the method that will best serve theoretical perspective / for better understand phenomenon or process
- มีจุดอ่อนที่คล้ายกับสองวิธีก่อนหน้า
- Has little guidance on how to use
- Unclear how to move between analysis
4. Concurrent Triangulation p. 217
- Use diff. method to
(1) Confirm
(2) Cross-validate
(3) Corroborate findings
- Method
(1) Use separate quan & qual.
(2) Offset weakness with strength of other method
(3) Quan / qual. Data collect concurrent (Equal priority)
(4) Integrate results during interpretation
(5) ใช้เวลาในการเก็บข้อมูลน้อยกว่า แบบ sequence
- ข้อจำกัด
(1) Require ความพยายาม, ความเชี่ยวชาญในการศึกษาปรากฏการณ์โดยอาศัยทั้ง 2 วิธีการ
(2) ยากในการเปรียบเทียบข้อมูล
(3) Unclear how to resolve discrepancy (ความไม่ลงรอยกัน / ความขัดแย้งในผลการศึกษา)

5. Concurrent Nested (multilevel design) p. 218
- Quan. / qual. Data collect simultaneous
- แต่มีวิธีการหนึ่งเป็น predominant ที่guide project โดยมีอีกวิธีแฝงอยู่
- Mixed data at analysis phase
- May or may not have guiding theoretical perspective
- Purpose:
(1) Broader perspective than using one method alone
(2) Use to describe aspect ที่ไม่สามารถนับเชิงปริมาณได้
(3) Utilize diff. method to study diff. groups or levels เช่น ลูกจ้างเก็บข้อมูลโดยใช้วีการเชิงปริมาณ ส่วนนายจ้าง / ผู้จัดการใช้วิธีการเชิงคุณภาพ
(4) Study each of treatment condition
- ข้อจำกัด
(1) ปัญหาเรื่อง discrepancy
(2) Unequal evidence
6. Concurrent Transformative p. 219
- Guided by theoretical perspective
- Base on ideologies: critical theory, advocacy, participatory research or conceptual framework
- สามารถใช้วีการทั้งของ nested…divert participants ..change process / triangulation…converge information to provide evidence for …
- Collect data same time
- May have equal / unequal priority
- Integrate at analysis phase
Data collection procedure
1. Discuss the specific types of data to be collected
- differ in terms of open-ended VS close-ended
- some form of data (interview / observation) can be either quan. or qual.
- reducing information to numbers use in quan. and qual.
2. Identify sampling strategies
2.1 Quan.
- involve random sampling
- equal probability of being selected
- can be generalized to large pop.
2.2 qual.
- purposeful sampling - selected because have experienced in phenomenon
3. Establish validity of data

Data analysis and validation procedures p. 220
- Data analysis in mixed method relates to types of research strategy chosen
- Procedures need to be identified within the design
- Analysis occurs both quan (descriptive and inferential numeric analysis) and qual. (description and thematic text or image analysis)
1. Data transformation…
1.1 creating codes & theme
1.2 counting the number of times they occur (in text data or talk)
1.3 quantify the qualitative data…enable to compare quan results with qual. Data
1.4 qualify quantitative data
by factors analysis..
- create factors or themes
- and compare with theme from qual. data
2. Explore outliers
2.1 extreme or outlier cases
2.2 why they diverged
3. Instrument development
3.1 initial qual
- obtain themes
- specific statement
3.2 use statement and theme for scales
- create survey instrument
3.3 validate instrument with large sample
4. Examine multiple levels (concurrent nested model)



Validity
- check validity both quan. and accuracy of qual.
- Triangulation:
(1) data source
(2) member-checking
(3) detailed description
(4) other approach
Report Presentation Structure (some guidance) p. 222
1. Sequential Study
1.1 quan. followed by qual.
- comments on how qual. Findings help elaborate on / extend quan results
1.2 qual. Followed by quan.
- present as two distinct phases
- separate headings for each phases
2. Concurrent Study
2.1 quan / qual may be presented in separate sections
2.2 analysis / interpret combine two forms to seek convergence among results
3. Transformative Study
3.1 structure involve advancing the advocacy issue (in beginning of study)
3.2 then use sequential or concurrent as a means of organizing the content
3.3 In the end: separate section may advance an agenda for change or reform (as a result of research)