Methodology is not a glossary of research terms. It is the reasoned explanation of how credible evidence will be generated, why a particular design is appropriate, and what limitations remain after those choices are made.
01 · WHAT DETERMINES YOUR METHODOLOGY?
Methodological choices should follow the study, not precede it.
A research topic alone does not determine whether a study is quantitative, qualitative or mixed methods. The appropriate methodological direction emerges from the research problem, research question, purpose, evidence required, degree of researcher control, time dimension, setting, unit of analysis, ethical feasibility and practical constraints.
Research problem→Question→Purpose→Evidence needed→Methodological approach→Design
Methodological determinant
Question to ask
What it helps determine
Nature of the problem
Is the issue about magnitude, relationship, meaning, experience or process?
Broad methodological direction
Research question
What exactly must be answered?
Quantitative, qualitative or mixed
Purpose
Describe, explore, explain, compare, predict or evaluate?
Research design
Evidence required
Numbers, narratives, observations, documents, digital interaction or combinations?
Data strategy
Researcher intervention
Will anything be deliberately changed or introduced?
Observational vs experimental
Time dimension
One point, repeated observations, before/after or historical?
Cross-sectional, longitudinal or time-series design
Unit of analysis
Individuals, households, organisations, documents, communities or online groups?
Design and sampling
Setting
Physical, institutional, community, archival or digital?
Possible methodological family
Existing evidence
Is suitable secondary data already available?
Primary vs secondary analysis
Feasibility
Can the design be completed ethically with available time, expertise and resources?
Practical design choice
02 · FIRST DECISION
Quantitative, qualitative or mixed methods?
Quantitative
Usually appropriate when the study asks how many, how much, how often, whether groups differ, whether variables are associated, what predicts an outcome, or whether an intervention produces measurable change.
Qualitative
Usually appropriate when the study asks how people experience something, what it means to them, why a process operates as it does, or how practices, identities and meanings are constructed.
Mixed methods
Useful when the study needs both a measurable pattern and a contextual explanation, and when the quantitative and qualitative strands are deliberately integrated.
03 · SAME TOPIC, DIFFERENT METHODS
The question changes the methodology.
Topic: Social media use among university students.
Research question
Evidence needed
Likely direction
What relationship exists between daily social-media use and academic performance?
Measurable exposure, measurable outcome, statistical association
Quantitative → cross-sectional/correlational
How do students describe the influence of social media on their study routines?
Experiences, meanings, context
Qualitative → interviews/focus groups
To what extent is social-media use associated with performance, and how do students explain the influence?
Statistical pattern + contextual explanation
Mixed methods
04 · PURPOSE OF THE STUDY
What is the study trying to accomplish?
Purpose
Typical question
Possible direction
Descriptive
What proportion of SMEs use mobile-money services?
Quantitative descriptive survey
Exploratory
How do informal traders experience the transition to digital payments?
Qualitative exploratory study
Relational
Is supervisory support associated with intention to leave?
Quantitative correlational design
Predictive
Which organisational factors predict employee turnover intention?
Quantitative regression-based design
Evaluative
Does a structured training programme improve knowledge scores?
Experimental or quasi-experimental
Explanatory
Why do some rural entrepreneurs abandon mobile banking after adoption?
Qualitative or mixed methods depending on evidence required
05 · QUANTITATIVE RESEARCH
Measurement, comparison, association and prediction.
Quantitative research is not simply “research with numbers.” Its logic is that concepts relevant to the research question are operationalised into measurable observations that can be analysed statistically.
Design
Main purpose
Example
Descriptive survey
Describe characteristics or prevalence
How many teachers use AI tools?
Cross-sectional
Measure variables at one point
Is workload associated with burnout?
Correlational
Examine statistical relationships
Is supervisory support related to retention?
Cohort
Follow groups through time
Does exposure predict a later outcome?
Case-control
Compare prior exposure between cases and controls
Was exposure more common among cases?
Longitudinal
Examine change over time
How does engagement change during postgraduate study?
Randomised experiment
Test an intervention under random allocation
Does training improve performance?
Quasi-experimental
Evaluate intervention without full randomisation
Did a policy change improve outcomes?
Interrupted time series
Test whether an intervention disrupted an established trend
Did service utilisation change after a funding reduction?
06 · QUALITATIVE RESEARCH
Understand meaning, experience, process and context.
Approach
Main focus
Phenomenology
Lived experience
Case study
Intensive examination of a bounded case
Ethnography
Culture, behaviour and social practices
Netnography
Online cultures and digital communities
Grounded theory
Developing theory from data
Narrative research
Stories and personal accounts
Participatory research
Research conducted collaboratively with participants
Action research
Studying and improving practice simultaneously
Discourse analysis
Language, meaning and social power
Qualitative content analysis
Patterns within textual or visual material
Reporting standards such as SRQR and COREQ help researchers report qualitative sampling, researcher roles, data collection and analytic procedures transparently.
07 · NETNOGRAPHY & DIGITAL RESEARCH
When the research setting is online.
Netnography adapts ethnographic principles to digital communities and cultures. It is useful where interaction, identity, trust, norms or meaning are constructed primarily through online environments such as WhatsApp communities, Facebook groups, Reddit communities, gaming spaces or professional networks.
Example question
How do young entrepreneurs construct trust and legitimacy within informal WhatsApp business communities?
Potential evidence
Ethical issue to consider
Posts, comments, shared images, interaction patterns, recurring language
Public versus private expectations
Community norms and digital practices
Consent and traceability of quotations
Researcher observation and field notes
Usernames, screenshots and platform rules
Netnography should not be reduced to “collecting social-media posts.” The method requires attention to culture, interaction, researcher positioning and ethics.
08 · MIXED METHODS
Integration is what makes a study mixed.
Design
Sequence
Typical purpose
Convergent
QUAN + QUAL
Obtain complementary views at roughly the same stage
Explanatory sequential
QUAN → QUAL
Use qualitative evidence to explain quantitative findings
Exploratory sequential
QUAL → QUAN
Use qualitative findings to develop concepts, measures or a later quantitative phase
The researcher observes conditions as they exist. Examples include cross-sectional, cohort, case-control and secondary-data studies.
Experimental
The researcher deliberately introduces an intervention and, in a true randomised trial, allocates participants randomly.
Quasi-experimental
An intervention occurs, but full randomisation is not possible.
Time also matters. Cross-sectional designs measure within a limited period; longitudinal designs observe change over time; retrospective designs use prior records; before-and-after and interrupted time-series designs evaluate change around an intervention or event.
10 · UNIT OF ANALYSIS & EXISTING DATA
Who or what is actually being studied?
The unit of analysis may be individuals, households, organisations, communities, policies, documents, countries, social-media posts or online communities. A methodology should match that unit.
Example
“How is climate adaptation represented in Zimbabwean agricultural policy documents?” may be better answered through document, content or discourse analysis than through a participant questionnaire.
Researchers should also ask whether credible secondary data already exist. A study of monthly HIV service utilisation before and after a funding change, for example, may be better suited to secondary quantitative analysis and interrupted time series than to collecting a new convenience survey.
11 · SAMPLING
A large sample does not automatically make a study representative.
A university may have 8,000 students and a voluntary online survey may collect 1,500 responses. The sample may still overrepresent students who use social media frequently, have reliable internet or are particularly interested in the topic.
Questionnaire design should begin with the evidence required to answer the research question. If the study examines supervisory support, the researcher may need to consider availability, feedback, guidance, fairness or recognition before writing individual items.
Weak item
My supervisor is knowledgeable and supportive.
Stronger item
My supervisor provides useful guidance when I experience difficulties in my work.
Cognitive interviewing can help identify questions that researchers and respondents interpret differently before full deployment.
13 · RELIABILITY & VALIDITY
Consistency does not prove correct measurement.
Reliability concerns consistency. Validity concerns whether the interpretation made from the measurement is justified. A weighing scale that always adds five kilograms may produce consistent results while remaining inaccurate.
14 · QUALITATIVE ANALYSIS
Quotations are evidence, not the final analysis.
Raw evidence
Code
Category
Theme
Interpretation
“Sometimes I wait two weeks before my supervisor responds.”
Delayed responses
Communication delays
Fragmented supervisory support
Support is inconsistent across time.
“One meeting tells me to change it, the next tells me to change it back.”
Conflicting feedback
Inconsistent guidance
Fragmented supervisory support
Students experience unstable academic direction.
A transparent analytical trail is: raw evidence → code → category → theme → interpretation.
15 · STATISTICAL INTERPRETATION
Statistical significance is not practical significance.
If a study reports r = .08, p < .001, n = 4,000, the result is statistically significant but the observed correlation is weak. A complete interpretation considers effect size, confidence intervals, design, measurement quality, sample size and substantive importance.
FIELDWORK & REAL DATA COLLECTION TOOLKIT
Translate the methodology into a controlled field operation.
Real data collection introduces practical risks that may never appear clearly in a written methodology: inconsistent interviewing, weak consent procedures, broken skip logic, duplicate submissions, device failure, poor supervision and late discovery of data-quality problems.
Approved questionnaire, interview guide or observation tool
Consent process
Participant information and consent procedure
Sampling plan
Eligibility and selection rules
Enumerator manual
Instructions for administering each section
Digital form
Tested electronic questionnaire where applicable
Validation rules
Required fields, ranges, skip logic and consistency checks
Field schedule
Team, location and deployment plan
Escalation plan
Who handles ethical, safety and technical problems
Quality-control plan
Daily review, back-checks and anomaly detection
17 · ENUMERATOR TRAINING
Enumerator training is part of data quality.
Enumerators should understand the study purpose, eligibility rules, consent, confidentiality, each question, skip logic, neutral probing, refusals, device use and escalation procedures before interviewing real participants.
Study briefing→Instrument walkthrough→Role-play→Device practice→Mock interview→Field readiness
Unusually high productivity or very short interview duration does not prove fabrication, but it is a quality-control signal that should be investigated.
23 · BACK-CHECKS & END-OF-DAY REVIEW
Find field problems while they can still be corrected.
Back-checks
Verify a subset of interviews where appropriate: did the interview happen, was the participant eligible, and were major procedures followed?
Creswell, J. W., & Creswell, J. D. (2022). Research design: Qualitative, quantitative, and mixed methods approaches (6th ed.). SAGE Publications. Publisher page
Creswell, J. W., & Plano Clark, V. L. (2025). Designing and conducting mixed methods research (4th ed.). SAGE Publications. Publisher page
Kozinets, R. V. (2019). Netnography: The essential guide to qualitative social media research (3rd ed.). SAGE Publications.
O'Brien, B. C., Harris, I. B., Beckman, T. J., Reed, D. A., & Cook, D. A. (2014). Standards for reporting qualitative research: A synthesis of recommendations. Academic Medicine, 89(9), 1245–1251. SRQR
Tong, A., Sainsbury, P., & Craig, J. (2007). Consolidated criteria for reporting qualitative research (COREQ): A 32-item checklist for interviews and focus groups. International Journal for Quality in Health Care, 19(6), 349–357. COREQ
EQUATOR Network. (n.d.). Reporting guidelines for main study types. EQUATOR Network
Centers for Disease Control and Prevention. (n.d.). Collaborating Center for Questionnaire Design and Evaluation Research. CDC CCQDER
World Health Organization. (n.d.). WHO STEPS surveillance manuals and data collection resources. WHO STEPS
