CPUT RMD210S: Research Methods for Development Practitioners

CPUT’s RMD210S: Research Methods for Development Practitioners equips future development practitioners with the logic, tools, and ethics of applied research in real-world settings. The course emphasises how to move from a practical development problem to a researchable question, how to design credible methodologies, and how to collect and analyse data responsibly—often in complex South African communities and institutions. These exam notes focus on exam-relevant concepts, step-by-step methods, and application to development work within South Africa’s higher education, TVET, NGOs, municipalities, and community contexts.

Across the five main sections below, you’ll cover research paradigms, problem formulation and literature review, research design and sampling, data collection and measurement, analysis and reporting, and ethics/quality/rigour. Throughout, examples are framed to match typical development topics seen in South African practice: livelihoods, youth employment, service delivery, food security, inclusive education, sanitation, public health communication, and community-based monitoring.

Research Paradigms, Epistemology, and the Development Practitioner’s Research Role

Why Research Methods Matter in Development Practice

Development practitioners rarely get to work in “laboratory” conditions. They deal with:

Human behaviour (attitudes, motivation, trust)
Institutional constraints (budgets, staffing, bureaucracy)
Power dynamics (who speaks, who is heard, whose knowledge is legitimised)
Context variability (urban vs rural; formal vs informal settlements; language differences)

Research methods provide a way to make your inquiry systematic, defensible, and ethically grounded. In RMD210S, the goal is not only to “collect data,” but to ensure that the data you collect can credibly answer the question your development intervention needs.

In exam questions, lecturers often reward students who can:

Explain which paradigm the research aligns to (and why).
Link paradigm → design choices (methods, sampling, analysis).
Justify ethics and rigour given context and constraints.

Core Research Paradigms (How You Know What You Know)

A research paradigm is a worldview about:

What reality is like (ontology)
How knowledge can be obtained (epistemology)
What counts as valid evidence (methodology)

The three most common paradigms you’ll see in development research are:

1) Positivism / Post-positivism

Assumption: Social phenomena can be measured and analysed using methods similar to natural sciences.
Ontology: There is a “real” external world that exists regardless of the observer.
Epistemology: Objective measurement is possible; researchers should minimise bias.
Typical in: surveys, experiments/quasi-experiments, quantitative impact evaluation.

Example: A development project introduces a youth entrepreneurship training programme. Positivist evaluation might measure changes in employment status, income, or business survival rates using a structured questionnaire and statistical analysis.

2) Interpretivism / Constructivism

Assumption: Social reality is constructed through meanings, experiences, and interactions.
Ontology: Multiple realities exist depending on people’s interpretations.
Epistemology: Knowledge is co-created with participants; understanding meanings is central.
Typical in: interviews, focus groups, ethnographic approaches.

Example: An interpretivist study explores why young people participate (or do not participate) in community training. The key data are narratives about barriers, aspirations, stigma, or trust in institutions.

3) Critical Theory / Critical Realism

Assumption: Social structures and power relations shape outcomes; “facts” alone do not reveal injustice.
Ontology: Underlying structures exist (often invisibly), but can only be understood through historical and social analysis.
Epistemology: Researchers should interrogate power, ideology, and inequality.
Typical in: participatory research, action research, political economy-informed studies.

Example: A critical study examines how gender inequality affects access to water services. Beyond measuring water availability, the analysis links outcomes to social norms, labour division, and unequal decision-making power.

Paradigm Choice Drives Method Choice

In RMD210S, it’s important to show you understand the relationship between paradigm and methods.

Paradigm	Common Questions	Typical Methods	Common Sampling Logic	Typical Analysis
Positivist/Post-positivist	“How much? How many? Does X cause Y?”	Surveys; quasi-experiments	Probability sampling; comparison groups	Quantitative statistics; regression
Interpretivist	“How do people make sense of X?”	Interviews; focus groups	Purposive sampling; maximum variation	Thematic analysis; grounded theory
Critical	“Who benefits? What power blocks change?”	Participatory research; case studies	Purposive; politically relevant groups	Thematic + structural interpretation; discourse analysis

Exam tip: When asked “Which methodology is most appropriate?” don’t reply only with a method name. State the paradigm logic and then justify the method.

Development Practitioner Research: Purpose and Constraints

Development work has distinct research purposes:

Needs assessment: Determine local priorities (e.g., water, sanitation, skills).
Feasibility and acceptability: Test whether an intervention is workable.
Baseline and monitoring: Establish starting points and track progress.
Impact evaluation: Determine whether outcomes changed because of the programme.
Process evaluation: Understand how implementation affected outcomes.
Learning and adaptation: Improve the next cycle of programming.

The constraints include:

Limited time (fieldwork windows; seasonal effects)
Language diversity and literacy levels
Safety considerations and community gatekeeping
Data availability issues (absence of records; inconsistent administrative data)

Therefore, method selection must be pragmatic but rigorous.

A Practical Example: Youth Skills Programme

Suppose a municipality plans a youth skills programme targeting unemployed young adults (18–35). A complete research plan could include:

Baseline survey (post-positivist): employment status, income sources, skills baseline.
Focus group discussions (interpretivist): perceptions of skills training, barriers, trust in organisers.
Stakeholder interviews (critical/interpretive): power relations between municipality, training providers, and communities.
Mixed-method impact evaluation (pragmatic): quantify job outcomes and explain why outcomes differ among subgroups.

This “one project—multiple research needs” logic is common in RMD210S exam questions: you should be able to propose research designs aligned to each purpose.

Research Ethics as Part of Knowledge Production

Ethics is not an add-on. Ethical principles affect:

Sampling choices (who you can recruit)
Consent methods (especially where literacy varies)
Data handling (confidentiality and anonymity)
The right to withdraw without penalty
The potential harm of revealing sensitive information (e.g., undocumented status, domestic violence, stigma)

Development research often engages vulnerable groups, which heightens ethical requirements.

Quality and Credibility in Different Paradigms

Students sometimes think “rigour” only applies to quantitative research. In reality, each paradigm has its own credibility criteria:

Positivist: reliability, validity, statistical power, measurement accuracy.
Interpretivist: credibility, transferability, confirmability, dependability.
Critical: reflexivity, emancipatory intent, attention to power and context.

In exams, you can score marks by stating which rigour criterion suits the approach and how you ensure it.

Problem Formulation, Literature Review, and Developing a Research Question

From Development Problem to Researchable Question

A development problem is usually broad: “poverty,” “low school performance,” “poor sanitation,” “unemployment,” “low uptake of maternal health services.” Research transforms a broad issue into a focused, investigable question.

A good research question typically:

Specifies the population (who?)
Specifies the phenomenon (what?)
Specifies the context (where? in what setting?)
Indicates the time frame (if relevant)
Can be answered using feasible methods and data

Example: Broad Problem → Research Question

Broad: “Youth unemployment is high.”
Researchable: “What barriers prevent young adults in Khayelitsha (Cape Town) from starting income-generating activities after attending a municipal entrepreneurship workshop?”

Now you can design:

a sampling strategy for workshop attendees/non-attendees,
an interview guide for barriers,
and potentially a survey for measurable outcomes.

Problem Statement: What Markers Want to See

In exams, a problem statement should communicate:

The practical development issue (what is happening?)
The gap in knowledge or practice (what do we not understand well enough to act?)
The expected contribution (why should the research matter?)
The scope (where, who, what boundaries)

A weak problem statement often describes only background without a clear research gap. A strong one explains why existing information is insufficient or outdated.

Research Objectives vs Research Questions

Often, exam questions distinguish between objectives (what you aim to achieve) and questions (what you ask).

Research question: guides inquiry directly.
Research objective: operationalises the question into achievable tasks.

Common Objective Types

Descriptive: identify patterns (e.g., “describe factors influencing…”)
Explanatory/associational: determine relationships (e.g., “examine the association between…”)
Exploratory: understand meanings (e.g., “explore how participants perceive…”)
Evaluative: assess programme performance (e.g., “evaluate the effectiveness of…”)
Predictive: forecast outcomes (often quantitative)

Building Variables and Concepts

For quantitative work, you typically operationalise variables such as:

outcome variable: employment status, uptake rate, service satisfaction
predictor variables: training attendance, age, education level, household assets
control variables: gender, prior experience, location type

For qualitative work, instead of “variables” you focus on concepts such as:

trust in authorities
perceived safety
stigma
empowerment
agency
cultural norms

In RMD210S, you should be able to switch language depending on method and paradigm, but demonstrate conceptual clarity either way.

Conceptual Frameworks: Connecting Ideas to Data

A conceptual framework is a map of how you think factors relate. It helps you:

structure your literature review,
develop research questions and sub-questions,
design instruments (questionnaires/interview guides),
interpret findings.

Example: Conceptual Framework for Sanitation Uptake

Institutional factors: service reliability, subsidy availability
Household factors: water access, household income, tenure status
Social factors: community norms, health beliefs
Individual factors: perceived susceptibility and benefits
Outcome: sanitation facility adoption/use

Even in qualitative work, a light conceptual framework can guide interview topics and analysis.

Literature Review: Not a Summary—A Synthesis

A literature review is not only “what others said.” It should:

Identify major themes and debates
Show how prior studies measured or conceptualised the topic
Identify gaps (population gaps, geographic gaps, methodological gaps)
Build your justification for the approach you choose

Typical Literature Review Structure

Broad context: policy frameworks and national statistics
Key themes: e.g., determinants, implementation barriers
Methodological patterns: common designs, limitations
Gap statement: what remains unclear and why your study matters

Research Gap: How to Phrase It

A gap statement often sounds like:

“While studies have measured X, few have examined Y in Z setting.”
“Most evaluations used quantitative methods without exploring lived experiences.”
“Evidence exists for urban areas, but not for rural communities with different service delivery mechanisms.”

In South African development contexts, geographic and institutional gaps are common:

Different provinces have different policy implementation patterns.
Rural infrastructure limitations change access.
Language and cultural factors affect participation and response quality.

Hypotheses (When Appropriate)

Not every study needs hypotheses. Hypotheses are typically used in quantitative designs where you predict relationships.

A hypothesis might be:

H1: Young adults who attend more training sessions are more likely to start income-generating activities within six months.
H0: There is no significant association between training attendance frequency and starting income-generating activities.

In qualitative studies, instead of hypotheses you might use guiding questions and propositions.

Developing Sub-Questions for Precision

Often a main question is too broad. Sub-questions sharpen focus. Example:

Main question:

“What factors influence uptake of a community health programme in Khayelitsha?”

Sub-questions:

“How do residents perceive benefits and risks of the programme?”
“What role does family influence play?”
“How do language and communication channels affect participation?”
“What trust issues exist regarding providers?”

This structure also helps during analysis because each sub-question becomes a theme or analytic category.

Aligning the Research Question with Methodology

Alignment is a core RMD210S skill. If your question is about meanings and experiences, a purely survey design may miss the depth. If your question is about “how many” or “how much change,” relying only on interviews may be insufficient.

Alignment checklist (exam-friendly):

Is the question measurable/answerable?
Does the research design match the question type (descriptive/explanatory/exploratory)?
Are the variables/themes defined?
Is the sampling plan appropriate for capturing those dimensions?
Can ethical permissions be realistically obtained?

Example: Translating a Question into Objectives

Research question:

“How do women informal traders in Mitchells Plain perceive barriers to accessing municipal trading permits, and how does this perception affect compliance?”

Objectives:

Describe the perceived barriers (administrative, financial, and social).
Explore how those perceptions influence decisions to apply or not apply.
Identify coping strategies used to deal with permit requirements.
Recommend practical improvements to the permit access process.

Research Design, Sampling, and Fieldwork Planning (Including South African Contexts)

Research Designs: Choosing the Right “Blueprint”

A research design is the overall plan for how you will answer the research question. It includes:

approach (quantitative/qualitative/mixed/multi-phase)
data sources
timeline and procedures
measurement strategy (if quantitative)
fieldwork plan (if qualitative)

Common designs in development research include:

Cross-sectional: collect data once at a point in time
Longitudinal: collect data across multiple times
Case study: deep investigation of a bounded system
Comparative study: compare groups/settings
Quasi-experimental: compare intervention vs comparison without full randomisation
Experimental: randomised control (less common in field programmes but possible)
Participatory/action research: cyclical involvement to improve practice

Mixed-Methods Research: Why and How

Mixed-methods combines strengths of different approaches. In development, it’s common because:

you need numbers for scale/coverage,
and narratives for understanding mechanisms and barriers.

A typical mixed-methods logic:

Quantitative component identifies patterns (e.g., low uptake).
Qualitative component explains why (e.g., trust, perceived stigma).
Findings inform programme adaptation.

Mixed-Methods Designs (Conceptual)

Sequential explanatory: survey first, interviews after to explain results.
Sequential exploratory: interviews first, survey instrument developed after.
Concurrent triangulation: collect both at the same time and compare.
Embedded design: one method supports another (e.g., qualitative embedded in a survey study).

In exams, when asked “why mixed-methods?” answer with complementarity: one method addresses what the other can’t.

Sampling: Strategy for Representativeness or Depth

Sampling answers: Who will you study and how will you select them?

1) Probability Sampling (Quantitative emphasis)

Used when you aim for representativeness and can compute sampling probability.

Key types:

Simple random sampling
Systematic sampling
Stratified sampling (divide into strata like gender, age, province)
Cluster sampling (select clusters then individuals within clusters)

Example: For a baseline survey across multiple wards, cluster sampling can reduce cost:

Select wards (clusters).
Within each selected ward, sample households.

2) Non-probability Sampling (Qualitative emphasis)

Used when the goal is depth, variation, and insight.

Common types:

Purposive sampling (choose participants relevant to the topic)
Maximum variation sampling (capture wide diversity)
Snowball sampling (participants refer others)
Convenience sampling (often weaker for inference but sometimes used due to constraints)

Example: If investigating experiences of informal traders facing permit barriers, purposive sampling selects:

traders who have applied but failed,
traders who applied successfully,
traders who never applied due to perceived barriers.

Sample Size: “Enough for Meaning,” Not Just a Number

Quantitative sample size depends on:

population size,
desired confidence level,
margin of error,
expected effect size,
variability.

Qualitative “sample size” depends on:

data saturation,
variation,
depth of inquiry.

A key exam principle:

For qualitative studies, you justify sample size through saturation and information richness.
For quantitative studies, you justify statistically or at least with power/precision logic.

Operationalising “Population” and “Unit of Analysis”

In exams, confusion often arises between:

population (who exists in theory),
sample (who you actually study),
unit of analysis (what entity you analyse).

Examples:

If studying “households,” units are households.
If measuring “individual perceptions,” units are individuals.
If analysing “school performance,” unit might be schools or learners depending on dataset.

In mixed-methods, unit of analysis must be consistent within each component.

Fieldwork Planning: Practical Steps

A strong research plan includes fieldwork logistics and timeline.

Fieldwork Steps

Permissions: university approval, ethics clearance, gatekeeper consent (e.g., municipal office, school principal).
Recruitment: lists, community announcements, referrals.
Training (if research assistants): how to approach participants, consent procedures, handling distress.
Pilot testing: test questionnaire/interview guide for clarity and cultural appropriateness.
Data collection: interviews, surveys, observation notes.
Quality control: daily debriefing, checking completed questionnaires, verifying recordings/transcriptions.
Data cleaning and secure storage: anonymisation, encryption, controlled access.

Pilot Study: Small Test, Big Value

A pilot study helps you detect:

unclear questions,
wrong language level,
leading questions,
missing response options,
time overruns,
misunderstandings.

In development contexts with diverse literacy, piloting is especially important for:

consent forms,
Likert scales,
questions about sensitive matters.

Data Collection Context in South Africa

When working in South Africa, field realities include:

official languages and local language preferences,
varying literacy rates and need for interviewer-administered questionnaires,
social desirability bias (participants may give “expected” answers),
community authority structures (ward councillors, traditional leaders, school governing bodies).

RMD210S expects you to demonstrate cultural sensitivity and procedural professionalism.

Example Fieldwork Plan: Water Service and Hygiene Promotion Study

Scenario: A project evaluates hygiene promotion impact in a Cape Town community.

Possible design:

Cross-sectional survey plus qualitative interviews.
Survey assesses:
- frequency of handwashing,
- perceived knowledge of hygiene,
- attitudes towards health messaging.
Interviews explore:
- how people interpret messages,
- why behaviour changes (or not),
- role of household decision-makers.

Sampling:

Stratify by residential area type (informal vs formal).
Within each area, cluster sample streets/blocks and then households.

Timeline:

Week 1–2: ethics permissions and recruitment plan.
Week 3: pilot test questionnaire.
Week 4–6: main survey data collection.
Week 7–8: interviews (purposive selection based on survey patterns).

This structure shows a coherent sequence from permissions to data collection.

Threats to Validity and How Design Helps Reduce Them

Validity threats:

Selection bias: who you recruit differs systematically from those you didn’t.
Measurement bias: instrument does not capture what it intends.
Confounding: other factors cause the outcome rather than the intervention.
Non-response bias: those who refuse differ in meaningful ways.
Recall bias: participants misremember events.

Design responses:

use appropriate sampling,
pilot instruments,
standardise interviewer training,
create comparison groups where feasible,
use triangulation across data sources.

Reflexivity and Positionality (Especially for Qualitative Work)

Reflexivity means reflecting on how researcher identity and assumptions shape data collection and interpretation.

In South Africa, reflexivity can involve:

language match/mismatch between researcher and participants,
power relations (e.g., researcher perceived as “from outside”),
prior relationships to community programmes,
professional status (government-linked vs NGO-linked).

In exams, you don’t need personal autobiography, but you must show awareness that researcher effects exist and can be mitigated through:

transparent procedures,
consistent interviewing practices,
reflexive journaling.

Data Collection, Measurement, and Data Management (Tools, Quality, and Practical Techniques)

Data Types in Development Research

You’ll encounter:

Quantitative data: numerical responses, scales, counts, rates.
Qualitative data: transcripts, field notes, documents, observations.
Mixed data: combined numeric and narrative evidence.

You must also manage data types differently:

Quantitative requires coding, cleaning, statistical assumptions.
Qualitative requires transcription, coding frameworks, and interpretation.

Instruments: Questionnaires, Interview Guides, Observation Protocols

A well-designed instrument ensures your research question can be answered.

Questionnaires

Typically include:

structured closed-ended questions (multiple choice, Likert scale)
demographic questions
sometimes open-ended “explain” responses

Key considerations:

Language clarity: use accessible phrasing.
Response options completeness: ensure options cover realities.
Avoid leading statements: minimise bias.
Order effects: decide question order carefully.

Interview Guides

For semi-structured interviews:

start with broad questions (warm-up),
move to specific topics aligned to objectives,
allow probing for depth.

A typical structure:

Context: participant background and experience
Perceptions: meaning, beliefs, motivations
Experiences: what happened, how, when
Barriers and facilitators
Recommendations for improvement

Focus Group Discussion (FGD) Guides

FGDs explore shared norms and differences. Moderator skills are critical:

manage dominant voices,
ensure respectful participation,
use prompts and interactive activities when appropriate.

Measurement and Operationalisation

Measurement means turning concepts into observable indicators.

Validity (Does it measure what it claims?)

Common types:

Content validity: items cover the full domain of the concept.
Construct validity: the instrument behaves as theory predicts.
Criterion validity: relates to an external standard (rare in some development settings).

Reliability (Does it produce consistent results?)

For scales: internal consistency (e.g., Cronbach’s alpha in many contexts).
For interviews: consistency in probing and coding.

In exams, you should mention:

pilot testing for face/content validity,
careful training for consistent data collection.

Scaling and Response Formats

Common options:

Likert scales: e.g., strongly agree → strongly disagree.
Frequency scales: never → always.
Agreement/importance scales: helpfulness, perceived effectiveness.

Pitfalls:

“Agreeing” may reflect politeness rather than true belief (social desirability bias).
Some participants interpret scale points differently.

Mitigation:

use interviewer explanations,
provide examples,
maintain consistent reading of questions.

Sampling Recruitment for Data Collection

Recruitment can introduce bias if not handled carefully.

Strategies:

approach participants systematically using sampling lists,
avoid recruiting only those who are easiest to access,
track non-response reasons (e.g., not home, refusals).

In exams, describe recruitment procedures clearly and link them to sampling logic.

Consent Procedures: Informed, Voluntary, and Ongoing

Informed consent typically includes:

purpose of study,
what participation involves,
risks and discomfort,
confidentiality and data handling,
voluntary nature and right to withdraw,
contact details for ethics or research team.

For low-literacy settings:

consent may be oral with a witnessed explanation,
use plain language and allow time for questions.

For vulnerable groups:

ensure safeguarding procedures,
consider additional permissions (institutional gatekeepers),
avoid coercion (especially if participants receive services).

Confidentiality and Anonymity

Confidentiality: protect identity and data access.
Anonymity: remove identifying details in reporting (e.g., no names, no specific addresses).

Practical methods:

assign participant codes (e.g., P001, P002)
store consent forms separately from responses
limit access to datasets
encrypt digital files or store on password-protected devices

Managing Data Quality During Collection

Quality checks:

field supervision and spot checks,
re-contact small subsets to verify unclear responses (if ethics permits),
daily review of completed questionnaires,
calibration of interviewers.

For qualitative data:

record with consent,
keep detailed field notes,
check audio quality,
debrief after each interview session.

Data Management: From Raw Data to Usable Data

A typical workflow:

Data coding (quantitative) or transcription (qualitative)
Data entry into SPSS/Excel/other tools
Data cleaning:
- check missing values,
- verify outliers,
- validate ranges (e.g., age must be within plausible bounds)
Documentation: codebook, variable definitions
Storage: secure backups and access control

In exams, you may be asked to define a “codebook” or “data dictionary.” A codebook lists:

variable names,
labels,
coding scheme (e.g., 1=Yes, 0=No),
missing value codes (e.g., 99=don’t know).

Triangulation: Using Multiple Data Sources

Triangulation strengthens credibility by comparing:

data sources (interviews + surveys),
methods (observation + interviews),
researchers (different analysts),
theories (different interpretive lenses).

In mixed-methods development studies, triangulation is especially valuable because:

numbers might show patterns but not explain causes,
narratives provide mechanisms and context.

Biases and How to Handle Them

Common biases:

Sampling bias: not everyone has equal chance.
Social desirability bias: participants answer to please.
Interviewer bias: interviewer tone or wording influences responses.
Recall bias: memory errors.

Mitigation:

standardise interview training,
neutral prompts,
assure confidentiality,
use appropriate time frames (e.g., “in the last 3 months”).

Example Instrument Design: Measuring Perceived Service Quality

For a service delivery study, you might measure:

reliability (service occurs as promised),
responsiveness (helpful when problems occur),
empathy (respectful interactions),
assurance (competence and trustworthiness).

You could create a Likert scale with items like:

“The service team arrives on time.”
“I feel respected when I interact with service providers.”
“When there is a problem, the team responds quickly.”

Then test:

whether items hang together as a coherent scale (reliability),
whether they align with theoretical constructs (validity).

Documenting Fieldwork: Field Notes and Audit Trails

An audit trail is a transparent record of decisions made during research:

sampling changes,
instrument amendments after pilot,
coding decisions,
deviations from protocol and reasons.

In RMD210S exam responses, referencing audit trails and documentation signals academic rigour.

Handling Sensitive Topics and Referrals

If participants discuss sensitive issues (e.g., violence, child abuse, illegal activity), researchers must:

provide support information (referral pathways),
stop or pause the interview if distress occurs,
protect participant identities more strictly.

A practical approach:

include referral contacts approved by ethics committee (where available),
train interviewers to respond empathetically and safely.

Data Analysis, Interpretation, Reporting, and Ensuring Research Quality

Analysis Begins Before Data Collection (With a Plan)

Although analysis happens after data collection, planning happens earlier:

decide the coding scheme for qualitative data,
define variables and categories for quantitative data,
clarify which analyses match each objective/question.

A good analysis plan reduces the risk of:

collecting more data than you can analyse,
“fishing” without clear links to the research question,
inconsistent interpretation.

Quantitative Analysis: Typical Steps

A standard quantitative workflow:

Descriptive statistics
- frequencies and percentages for categorical data
- means/medians for numeric variables
Cross-tabulations (e.g., relationship between training attendance and employment status)
Inferential tests (depending on design)
- chi-square for categorical associations
- t-tests/ANOVA for mean differences
Regression analysis (if appropriate)
- logistic regression for binary outcomes
- linear regression for continuous outcomes

Descriptive Statistics: What They Tell You

Descriptives answer:

Who are the respondents?
What levels do outcomes take?
How common are specific barriers or behaviours?

Example: A survey on sanitation adoption might report:

proportion with access to toilets,
proportion who report consistent use,
distribution of perceived barriers.

Qualitative Analysis: Common Methods

Qualitative analysis includes systematic coding and interpretation. Common approaches:

Thematic analysis (identifying themes across interviews)
Grounded theory (iterative coding to build conceptual theory)
Content analysis (coding explicit content)
Narrative analysis (focus on story structure)

Thematic Analysis: A Practical Multi-Step Approach

familiarisation: read transcripts and field notes
initial coding: assign labels to meaningful segments
theme development: group codes into broader themes
review and refinement: ensure themes fit data
write-up: connect themes to research questions and literature

Exam note: You should describe phases clearly and mention how you move from codes → categories → themes.

Mixed-Methods Integration: How to Combine Findings

Integration is often where students lose marks—because they present results separately without explaining how they connect.

Integration strategies:

Compare quantitative patterns with qualitative explanations.
Use qualitative findings to interpret unexpected quantitative results.
Merge findings in a joint narrative around each research objective.

Example:

Quantitative results show low permit compliance among traders.
Qualitative interviews reveal that compliance attempts are blocked by unclear procedures and fear of harassment.
Integration: interpret numbers through mechanism explanation.

Developing a Codebook for Qualitative Analysis

A codebook is not just for quantitative. For qualitative:

define code names,
specify what counts as inclusion/exclusion,
provide example excerpts.

This improves consistency between coders and supports dependability.

Ensuring Credibility, Transferability, Dependability, Confirmability

Qualitative rigour criteria commonly align to:

Credibility: do findings reflect participants’ perspectives?
Transferability: can others judge applicability to other contexts?
Dependability: is the process consistent and transparent?
Confirmability: can others see how interpretations were reached?

Strategies:

member checking (where appropriate),
triangulation,
detailed thick description,
reflexive journaling,
audit trail.

Validity and Reliability in Quantitative Analysis

Quantitative rigour involves:

measurement validity (do items capture constructs?)
reliability (internal consistency, stable measurement)
assumption checks (for parametric tests)
missing data handling (describe approach)

In exams, you can be asked: “How do you ensure validity?” Provide steps like:

pilot testing,
expert review of instruments,
construct-based item design,
transparent operationalisation.

Dealing with Missing Data

Missing data is common in fieldwork. Approaches depend on the nature:

listwise deletion (exclude cases with missing variables)
imputation methods (if suitable)
treat “don’t know” as a separate category (if justified)

In exam responses, emphasise:

describe how missingness was treated,
justify why method was chosen,
avoid hiding missing data effects.

Interpretation: Moving from Results to Meaning

Interpretation should connect:

results to research objectives,
findings to literature,
possible explanations considering context and theory.

A strong interpretation includes:

what the results mean,
why they might have occurred,
what alternatives exist,
implications for practice and policy.

Discussion: Addressing Contradictory Findings

Development contexts are messy. Results may contradict expectations:

changes might be smaller than anticipated,
some groups may benefit more,
implementation barriers may reduce impact.

Good discussions:

acknowledge unexpected findings,
propose plausible reasons linked to theory and context,
compare with studies that support/contradict results,
explain limitations.

Limitations: What Lecturers Look For

Limitations are not confessions of failure; they are scholarly boundaries.

Common limitations:

sampling constraints due to access,
language translation issues,
self-report bias,
cross-sectional design cannot infer causality,
limited time for deep engagement.

In exams, you should also propose how limitations could be mitigated:

improved sampling,
longitudinal follow-up,
stronger instrument validation,
triangulation and better training.

Ethics Revisited in Reporting

Ethics affects reporting:

avoid identifying details,
anonymise quotes,
secure storage of datasets,
report in aggregated form when necessary.

For sensitive qualitative data, quoting verbatim may require careful anonymisation and redaction.

Reporting Structure: Making Findings Communicable

A typical research report includes:

abstract/summary
introduction and background
literature review
methodology (design, sampling, instruments)
results
discussion
conclusion and recommendations
references
appendices (instruments, consent forms, coding schemes)

In exams, you may be asked to outline report components or explain what must appear in the methodology section.

Example: Reporting Findings for a Programme Evaluation

Suppose a training programme aims to improve employment outcomes. Your results section might include:

baseline participant demographics,
changes over time in employment status (if longitudinal),
statistical association between attendance and outcomes,
qualitative themes explaining how participants experienced training.

Your discussion should:

interpret changes relative to theory and past studies,
identify implementation lessons,
link barriers to mechanisms (e.g., transport costs reducing attendance continuity).

Research Quality and Trustworthiness Summary

You can summarise research quality using a “triad” approach:

Design quality: alignment of question, method, sampling.
Data quality: instrument validity/reliability; fieldwork consistency.
Interpretation quality: transparent analysis; ethical reporting; acknowledgement of limitations.

Ethics, Reflexivity, Governance, and Academic Integrity in RMD210S Research Practice

Core Ethical Principles in Social Research

Ethical research aims to protect participants and ensure responsible knowledge generation. Common principles include:

Respect for persons
- dignity, voluntary participation, informed consent
Beneficence
- maximise benefits, minimise harm
Justice
- fair selection of participants; avoid exploiting vulnerable groups
Integrity
- honesty in data handling and reporting

In development contexts, these principles require careful translation into field realities.

Informed Consent: Practical Challenges

Informed consent can be complex:

participants may have limited literacy,
language barriers may lead to misunderstanding,
power dynamics can make “voluntary” appear forced.

Mitigations:

provide consent information in preferred language,
use trained interpreters (and ensure confidentiality),
allow time for questions,
ensure that refusal does not affect access to services.

Vulnerable Populations: Heightened Ethical Care

Vulnerability may arise due to:

age (children, adolescents),
disability,
poverty (economic dependency on institutions),
stigma (HIV status, mental health),
migration/refugee status.

Ethical steps include:

additional safeguards approved by ethics committee,
careful phrasing to avoid coercion,
safe referral pathways for distress.

In RMD210S exam responses, mentioning how you would protect participants often earns more marks than simply stating that you will be ethical.

Confidentiality in Small Communities

In small communities, anonymisation becomes difficult because:

unique circumstances make identification possible,
participants may be able to infer identities from details.

Mitigation:

remove specific identifiers,
generalise descriptions in reporting,
consider participant consent regarding quotations and voice recordings.

Data Protection and Storage

Ethics governance includes responsible data storage:

password-protected devices,
encrypted files,
locked physical storage for paper documents,
clear retention timelines and secure destruction where applicable.

If data is stored on cloud services, ensure compliance with institutional policies.

Reflexivity and Power: Researcher Influence on Data

Reflexivity acknowledges that knowledge is shaped by researcher choices and social positions.

Potential reflexive issues in South African fieldwork:

researcher affiliation (university vs NGO vs municipality) influences trust,
gender dynamics can affect who feels comfortable answering,
language proficiency affects rapport and data depth.

Reflexive practice:

field journals documenting assumptions and interaction patterns,
consistent questioning style,
debriefing sessions with supervisors or team members.

Ethical Considerations for Community Gatekeepers

Community gatekeepers may include:

ward councillors,
traditional leaders,
school principals,
clinic managers,
NGO coordinators.

Ethical tension:

gatekeepers may control access and influence who participates,
gatekeeper presence during consent may compromise perceived voluntariness.

Ethical strategy:

negotiate appropriate access,
obtain consent privately when feasible,
ensure participants can refuse without consequences linked to gatekeeper authority.

Research Governance: Ethics Committee and Approvals

Most South African universities require ethics clearance before data collection. Governance often includes:

ethics application forms describing design, sample, recruitment, risks, benefits,
consent documents,
instruments and recruitment scripts,
data management plan.

In exams, when asked “what ethical approvals are needed?” students should mention:

institutional ethics committee clearance,
permission from relevant institutions (schools, municipalities, clinics),
compliance with university policies and national research ethics norms.

Risk Assessment: Anticipating Harm

Risk assessment in development research includes:

psychological harm (discussing trauma),
social harm (reputation or community consequences),
physical risk (travel safety in fieldwork),
legal/administrative harm if data reveals sensitive issues.

A strong risk section includes:

the likelihood and severity of harm,
mitigations (trained interviewers, referral contacts, safe scheduling),
emergency procedures and reporting obligations if necessary.

Academic Integrity: Plagiarism, Citation, and Proper Use of Sources

Academic integrity is essential for ethical research reporting. RMD210S expects:

correct referencing (consistent citation style),
proper paraphrasing,
no fabrication of data,
no misrepresentation of sources.

If a student uses secondary data (e.g., official statistics), it must be cited and properly described. If using online materials, record the source and date accessed.

Data Fabrication and Misrepresentation: Why It’s Unethical and Unscientific

Fabrication includes:

inventing participant responses,
altering data without documentation,
misreporting methods.

Misrepresentation includes:

presenting convenience samples as probability samples,
claiming ethical consent where none exists,
hiding missing data or selective reporting.

In exams, a short answer that explains the ethical and methodological impact can gain marks: it undermines trust and invalidates findings.

Reporting Ethics: How to Avoid Harm Through Publication

Ethics continues after fieldwork:

avoid identifying communities in ways that could stigmatise them,
avoid linking quotes to specific demographic descriptors where it increases identification risk,
handle sensitive themes responsibly.

Example Ethical Scenario: Interviewing Caregivers about Child Health

Scenario: A researcher interviews caregivers about childhood illness experiences.

Ethical risks:

distress when discussing sick children,
risk of identification if community is small,
confidentiality when clinics are involved.

Ethical mitigations:

consent with clear explanation,
option to pause or stop interview,
anonymised transcripts and redacted quotes,
referral information for caregiver support where appropriate,
no sharing identifiable details with clinic staff beyond required safeguards.

Evaluation Ethics: Avoiding Undue Burden on Communities

Impact evaluation often involves:

questionnaires,
repeated follow-ups,
monitoring visits.

Communities may face survey fatigue. Ethical mitigation:

reduce unnecessary questions,
coordinate with existing programme activities,
schedule visits at convenient times,
keep participant interaction respectful and time-bound.

Reflexive Reporting: Being Transparent About Researcher Role

Ethically strong reports include transparency about:

researcher assumptions,
how consent and recruitment were implemented,
how data were analysed and interpreted.

Transparency strengthens credibility and reduces the risk of misleading claims.

Integrative Ethical Framework for Your Exam Answers

When answering ethics exam questions, a high-scoring structure is:

Identify the ethical issue (e.g., confidentiality, consent, vulnerability).
Explain why it matters in this context (power, stigma, risk).
Propose mitigation actions (specific procedures).
Explain how you will verify ethical compliance (monitoring, documentation).

This “issue → context → mitigation → verification” approach is clear, logical, and demonstrates practical competence.

Conclusion: Mastering RMD210S Exam Competence Through Method-Logic and Ethical Rigour

CPUT’s RMD210S: Research Methods for Development Practitioners tests whether you can do more than define concepts—you must demonstrate the ability to connect paradigms to designs, transform problems into research questions, select sampling strategies, collect data credibly, analyse systematically, and report ethically. In South African development contexts, these skills must also respond to language diversity, institutional realities, and community power dynamics.

To excel in the exam, practise answering with the following integrated model:

Start with the development problem → justify the research gap
Choose research questions aligned to objectives
Match paradigm → design → methods → sampling
Plan instrument and fieldwork with data quality
Analyse with appropriate rigour criteria
Report responsibly with ethics and transparency

That model captures the heart of RMD210S and prepares you for real development practice where evidence, ethics, and credibility are inseparable.