Week 3 Flashcards
Quantitative and qualitative research
characteristics
<table>
</table>
Strictly speaking, “qualitative” and “quantitative” are characteristics of data, not research!
Qualitative research:
–>Types of research: exploratory, theory-building
–>Methods: case studies, interviews
–>Analysis method: coding
–>Conceptual models as results
–> “Subjective” results
Quantitaive research:
–>Types of research: heory-testing, decision science
–>Methods: surveys, experiments, mathematical modeling
–>Analysis: statistics, econometrics
–>Effect sizes, hypothesis tests, optimal choices as results
–> “Objective” results
Combination of research strategies (quant & qual)
All types of strategies can be combined!!!
* ‘Qualitative’ methods can be combined (e.g., interviews + archival research)
* ‘Quantitative’ methods can be combined (e.g., survey + modeling)
* ‘Qualitative’ and ‘quantitative’ methods can be combined (e.g., interviews to explore a problem/ help formulate hypotheses + testing through survey or experiment)
Things to consider when combining quantitative and qualitative research
Mixed-methods research requires very different research skills and is time-consuming, however…
- One method (used first) can help in designing the following strategy better
- Using multiple methods can enrich interpretation of findings
- Better methodological quality (‘triangulation’) and better generalizability
Research strategies for “Qualitative” research
- Ethnography
- Case studies
- Focus groups (if used as a stand-alone research method*)
- Interviews (if used as a stand-alone research method*)
- Archival research (secondary data, or desk research*), when it concerns qualitative secondary data
!!Sometimes certain research fields have strong preference for a research strategy!!!
Interviews: advantages & disadvantages
remember biases!
(-)Information bias: gives opinions, not “accurate” information
(-)Recollection bias: depends on interviewee memory, not suitable to discuss what happened a long time ago
<specific>
(-)Researcher bias: subject to interviewer's interpretation
(-)may be less suitable for sensitive topics
(-)lots of time needed to do it
(+) Widely used as "easy to obtain" if easier than access to documents
--> if there are "non-obtrusive" ways better to do those than interviews.
</specific>
Interview types
- Unstructured
–>for theory building
(Used when you have no knowledge or theory and need to explore a topic)
*free flow chat
*no two are the same
*interviewee leads - Semi-structured
–>for theory-building and elaborating
(Used when need good insights with some structure)
*questions from theory or practice
*order of questions may change during
*additional or follow up questions can be asked - Structured
–>for theory-building, elaborating and testing.
(Used when you need to compare opinions)
*possibly many closed questions
*questions derived from theory or practice
*easy analysis and comparison
“Semi-structured interviews”
types of questions
- Open questions- mainly used
*yes/no answer NOT POSSIBLE
*only one question inside!!!
*start with question word - Closed questions- minimise use
*do not give details
*can be used to verify facts - Leading questions- must avoid
*leading to the answer that you expect
“Semi-structured interviews”
interview guide
- Start with a topic in mind
- Think about your potential respondents
- Divide the topic into some subtopics
- Develop questions for each subtopic
- Review the logic and flow
- Review the guide after a couple of interviews
Who to Interview & How Many?
*Interview knowledgeable informants with different perspectives (stakeholders, departments, interests).
*Adjust questions as needed and start with a preliminary list, but stay flexible.
*Use snowballing (ask interviewees for more suggestions).
*More interviews are better, but time constraints apply—complex topics need more.
*Aim for saturation (stop when no new insights emerge).
*Ensure triangulation: by asking the same questions to multiple people.
Explain case study as a research strategy
*Investigates a recent phenomenon in its real-life context, especially when boundaries between phenomenon and context are unclear.
*Uses multiple data sources (qualitative & quantitative, though often linked to qualitative research).
*Suitable for exploratory, theory-building, and theory-testing studies.
Advantages and disadvantages of case studies
(+) Deep dive into real-life context
(+) Possible to study a new phenomenon (even if there is no prior theory)
(+) Possible to uncover issues that are not easily identified through other methods
(+) Findings are typically well received by practitioners
(+) Very versatile
(-) Discretion of the researcher about which data to include (researcher bias risk)
(-) Time- and resource-intensive
(-) Lack of (statistical) generalizability
(-) Hard to study past phenomena due to recollection bias and lack of information
(-) Difficult to write about concisely but convincingly
Types of case studies
- Single case-studies
*Used for unique/extreme phenomena or when there’s strong data access.
*Provides rich insights but has limited generalizability.
*Cannot prove universal truths, but can highlight gaps in existing theories (prove the opposite). - Multiple (comparative) case studies
*Replicate & enrich data, increasing generalizability.
*More convincing than single case studies.
*complexity- case selection is key—should cases be similar, different, or both?
*How many? More is better for theory testing, but balance is needed (depth vs. number, time, and comparability).
Both case study types can be divided further into…
Choice always driven by RQ/RO:
1. Cross-sectional vs Longitudinal
Cross-sectional: time dimension is not important – you look at data and study concepts at one specific point of time
Longitudinal: time dimension is important – you study a concept over time; so multiple measurements over time take place
2. Real-time vs retrospective
Real-time: you study something as it is happening
Retrospective: you study something that is in the past
–>Combination of real-time and retrospective possible!
–>Common mistake: I am doing study now, so it is real-time –no!
Single-case studies more often longitudinal (lots of time needed).
Multiple-case studies often cross-sectional.
Multiple-cases are more often retrospective.
Key element in case studies: unit of analysis!
- Unit of analysis – what you need to look for (or sample) in your cases to answer your RQ
- In essence, it related to how you operationalize theoretical or complex concepts/constructs
Always linked to the RQ!
How to Sample Cases?
No random sampling—cases must provide relevant insights to you phenomenon (THEORETICAL SAMPLING).
Single case: Choose an extreme or data-rich case.
Multiple cases:
Similar cases → maximize replication.
Different cases → compare success vs. failure.
Mixed approach → balance similarity & dissimilarity.
Common data sources (types) in case studies
*Interviews
Main problem – they provide personal opinion or judgments, and are prone to recollection bias.
*Observations
Can complement interviews and help to interpret them – and vice versa
*Archival documentation
Useful for replicating past story
To verify info gained in interviews
Sometimes challenging to gain access
*Other: videos, photos, questionnaires, industry reports
Why case studies require more than one source of data?
*Ensures rich data and convincing results (at least two sources).
*Different sources serve different purposes (e.g., context, diverse perspectives).
*Data triangulation improves quality by cross-verifying information.
Example: Check interview claims against archival documents.
*Common sources: Interviews + archival analysis, but can include quantitative data too!
General quality criteria for research (qualitative data):
Validity types
Validity (three types)
1. Construct – is the operationalization of the construct ‘sound’?
2. Internal – is there a strong link between collected data and developed theoretical ideas (does the data support the findings/conclusions)?
3. External – to which degree can findings be generalized beyond the context?
General quality criteria for research (qualitative data):
Reliability
*Replication is difficult due to unique settings and cases.
*Thorough documentation is a proxy for reliability in qualitative studies.
*While results can’t be replicated, the research process can be recreated.
General threats to quality in qualitative research
Informant Bias (→ Construct Validity)
Influenced by values, opinions, memory, honesty, and social desirability.
More likely if:
*Topic is sensitive/personal (e.g., money, promotion).
*Interviewee has personal stakes (e.g., failed project manager).
*Event happened long ago (memory issues).
Researcher Bias (→ Internal Validity)
Happens when personal views influence case selection or data analysis.
Lack of research skills can cause poor study design and weak conclusions.
Idiosyncratic Findings (→ External Validity)
Results may be true for the sample but hard to generalize if cases are highly unique.
How to fix validity issues in case studies
<check slide 41 for the table>
What are the challenges in qualitative data analysis?
- No clearly defined standards or detailed rules
- Choice of the right method is not always obvious
- Often learned by doing or from other researchers, rare systematic training
- Hard to transparently explain and present
- Can be very time-taking and costly – as there are large volumes of rich data
- Based on the individual interpretation – researcher bias risk
- Risk of poor quality of analysis if researchers have no right skills
Steps of analysing qualitative data
- Collect
- Organize and prepare for coding (transcribe)
- Code
- Analyze for insights
* Describe, compare, relate - Report the insights
* Develop findings, supporting arguments, defend your arguments using data, extend the results beyond the original research setting (link back to the existing knowledge)
Not a linear process tho!!!
Approaches of qualitative analysis
- Deductive approach – when the study is informed by previous theory and concepts, so main themes/categories are known, and data needs to be categorized into them (top-down approach)
–>more structured
–>focus on existing theoretical concepts - Inductive approach – when there is no previous theory, so categories and themes need to emerge from data (bottom-up approach)
–>less structured
–>focus on interpretation - Abductive approach – some parts of data are informed by existing theory, but it is likely that not all data will ‘fit’ into these themes; these parts of data are novel and need to be coded inductively (start from deduction, continue with induction)
