Wits SOCL7045A: Masterclass Notes on Social Research Design and Methodology

Social research thrives on design choices that make evidence credible, arguments traceable, and ethics non-negotiable. SOCL7045A develops advanced capability to plan, justify, and execute research using rigorous methodological reasoning—spanning epistemology, sampling, measurement, fieldwork, analysis, and research ethics. These notes are framed for Wits Focus: Labour, Policy & Globalisation Studies, with a strong emphasis on research designs relevant to South African universities, colleges, and TVET contexts, particularly where labour and policy questions intersect with global dynamics.

1) Research Design as Argument: Epistemology, Ontology, and the “Why” Behind Methods

What “research design” means in advanced social research

A research design is not a checklist of methods. In SOCL7045A-level work, it is a coherent plan for producing defensible knowledge. Your design functions as an argument that answers at least four questions:

What do you claim to know? (the research claim)
What is reality like? (ontology)
How can knowledge about that reality be produced? (epistemology)
What methods and procedures will generate credible evidence for that claim? (methodological strategy)

A design is therefore “methodological” and “theoretical” at the same time: it links conceptual commitments to empirical procedures.

Example (labour policy and globalisation):
Suppose you want to explain how trade union bargaining strategies change under global supply-chain pressure. A design that relies only on cross-sectional surveys might measure “perceptions of pressure” but struggle to explain “how strategies changed” over time. A design using process tracing or a mixed-methods longitudinal approach may align better with the “how” and “why” logic you intend to defend.

Epistemology: positivist, interpretivist, critical, and pragmatic positions

Advanced social research design often requires you to position yourself relative to philosophical traditions. Rather than memorising labels, focus on what these positions imply for evidence and inference.

Positivist-leaning approaches often prioritise measurement, comparability, and probabilistic inference.
- Strength: clarity about variables, causal models (when assumptions are credible).
- Risk: treating complex social processes as if they were uniform “objects.”
Interpretivist approaches emphasise meaning, social action, and context.
- Strength: capturing lived experience, local categories, and how actors interpret structures.
- Risk: under-explaining broader patterns if sampling and analytic strategies are weak.
Critical approaches highlight power, ideology, structural constraints, and the production of knowledge itself.
- Strength: sensitivity to domination and historical specificity; ethics as part of research design.
- Risk: can become too general if empirical grounding and operationalisation are vague.
Pragmatist positions emphasise using methods that work for the research problem.
- Strength: method choice is driven by the claim.
- Risk: “anything goes” if justification is not rigorous.

SOCL7045A expectation (typical): your design should show that your philosophical stance is not decorative—it informs how you define concepts, create instruments, sample participants, handle bias, and make claims.

Ontology: structure, agency, and relational realities

Ontological commitments determine what you treat as real and how you theorise causation or explanation.

Common ontological orientations include:

Structural orientations: social outcomes are shaped by institutions, labour markets, policy regimes, welfare systems, global trade rules.
Interpretive/constructivist orientations: outcomes are mediated by meanings, identities, discourses, and interactions.
Relational orientations: social realities emerge through relationships among actors, institutions, and material conditions.

Labour-policy example (South Africa):
If your research focuses on employment policy implementation in a district municipality or sectoral bargaining forum, you cannot assume that “policy” simply flows from national documents into workplaces unchanged. Implementation is mediated by street-level bureaucrats, union strategists, employer compliance capacity, and fiscal constraints. A relational ontology will encourage you to design fieldwork or document analysis that tracks these mediating relationships.

The logic of inference: describing, explaining, and generating mechanisms

Design at this level must clarify the type of contribution you aim for:

Descriptive: documenting patterns (e.g., how many workers in informal work receive labour protections).
Explanatory: accounting for variation or change (e.g., why compliance differs between provinces).
Mechanistic: showing how processes produce outcomes (e.g., how fear of retaliation shapes complaint reporting).
Evaluative: assessing effectiveness of interventions (e.g., policy outcomes compared with stated aims).
Interpretive-theoretical: refining concepts (e.g., what “decent work” means in a specific industry context).

Your design choices should match your inference goals. For example:

A descriptive question often fits with well-specified sampling and measurement.
A mechanistic question often needs temporal depth, sequence, and evidence triangulation (e.g., interviews plus organisational documents plus event timelines).

Research paradigms translated into concrete design features

To make design “real,” translate philosophical commitments into specific decisions.

Decision map (use in assignments):

Claim type → appropriate evidence
Ontology → the level of analysis and what counts as “causal”
Epistemology → how you handle validity, bias, reflexivity
Methods → operationalisation, sampling, data collection, analytic strategy
Ethics → participant risk management and governance

Micro-case:
If you argue that “policy implementation is shaped by institutional capacity and labour relations,” then design features might include:

purposive selection of sites with different institutional capacity (budget size, staff availability)
document analysis of policy communications and compliance reports
interviews with both state implementers and labour stakeholders
analytic tracing of how capacity constraints become concrete decisions (e.g., prioritisation of certain sectors, backlog management)

This is not just “using multiple methods”—it is designing for a specific kind of explanation.

2) Sampling, Measurement, and Operationalisation for Labour, Policy, and Globalisation Research

Sampling logic: probability, purposive, and theoretical sampling

Sampling is where your design becomes empirically credible. Advanced research requires you to justify why your sample is capable of answering the research question, not merely why it is convenient.

Probability sampling

Used when you need estimates representative of a defined population.

Common designs:

Simple random sampling
Stratified sampling (e.g., by province, industry, gender)
Cluster sampling (e.g., firms within municipalities)
Multi-stage sampling (e.g., households → individuals)

Key requirement: clear population boundaries. For South African studies, your “population” might be:

registered employers in a specific sector
employees in particular labour categories
workers in a district municipality’s catchment area

If the population is unclear, representativeness becomes impossible to defend.

Non-probability purposive sampling

Used when the goal is depth, relevance, or variation, not statistical generalisation.

Common strategies:

Maximum variation sampling (different types of unions, sectors, provinces)
Typical case sampling (average conditions)
Extreme/deviant case sampling (highest conflict, lowest compliance)
Expert sampling (policy makers, labour inspectors, union organisers)

Theoretical sampling (grounded theory logic)

Used iteratively as your analysis develops. You collect data to test developing concepts, refine categories, and strengthen the emerging theoretical account.

Labour-policy example:
If early interviews suggest that “informal retaliation fears” block complaint reporting, you might later sample workplaces where retaliation is known to be high versus workplaces where it is low, to clarify the concept’s boundaries.

Sample size: what matters more than a single number

In SOCL7045A contexts, it’s tempting to hunt for “the right sample size.” A better approach is to use a justification framework:

For quantitative surveys: precision and power
For qualitative studies: data saturation, conceptual completeness, variation coverage
For mixed methods: joint sufficiency for both components

Quantitative sample size reasoning

While exact power calculations depend on effect sizes and variance, you should always specify:

the outcome variable type (continuous, binary, ordinal)
expected effect or comparison contrast
target confidence/precision
design effects (clustering and weights)

If you claim 400 participants give 95% confidence:** ensure that the confidence claim matches your design and sampling plan.** In many real assignments, the “best” approach is to show a transparent calculation rather than using a universal rule.

Qualitative sample size reasoning

Qualitative sample size is often justified via:

saturation (no new themes emerge)
thick description (sufficient evidence for claims)
coverage of variation (different stakeholder positions and contexts)

For labour and policy research in South Africa, saturation often depends on:

number of provinces involved
industry diversity (mining vs retail vs public sector)
stakeholder diversity (workers, unions, employers, regulators)

Operationalisation: turning concepts into observable indicators

Operationalisation is a core method skill—especially when you study abstract constructs like “precarity,” “policy capacity,” or “globalisation pressure.”

Three levels of operationalisation

Conceptual definition
- What the concept means in your theoretical frame.
Dimensions
- What sub-aspects make up the concept.
Indicators
- What you will measure (survey items, interview themes, document variables).

Example concept: “Labour policy effectiveness”
You could define it as:

compliance with labour standards
reduced workplace incidents
improved grievance resolution times

Then select indicators:

inspection reports and recorded outcomes
time-to-resolution for complaints
worker-reported experiences of enforcement

You then ensure that indicators align with dimensions rather than mixing unrelated measures.

Measurement validity: content, construct, and criterion considerations

Measurement validity asks: does your tool actually measure what you claim?

Content validity: Do your items cover the full domain of the construct?
Construct validity: Do items behave like your theory predicts (convergent/discriminant patterns)?
Criterion validity: Do measures correlate with an external “gold standard” when available?

Labour studies frequent pitfalls:

Using “employment status” proxies as if they reflect job quality.
Confusing “access to social protection” with actual benefit receipt.
Treating “awareness of rights” as equivalent to “rights enforcement.”

To avoid these, specify whether your research measures:

awareness, ability, access, usage, and outcomes separately.

Reliability and consistency: ensuring evidence stability

Reliability matters even in qualitative research—though the framing differs.

In quantitative surveys: internal consistency (e.g., Cronbach’s alpha), test-retest reliability, inter-rater coding reliability.
In qualitative research: consistency of coding, transparent codebooks, reflexive memoing.

Practical method: develop a coding framework early and revise it systematically. Track changes and document how and why.

Managing measurement bias: social desirability, recall bias, and strategic responses

In labour and policy settings, bias risks are high:

Social desirability bias: participants may report what is “acceptable” to regulators or researchers.
Recall bias: workers may misremember timelines of incidents.
Strategic responses: employers or officials may frame compliance to avoid penalties.

Design strategies:

triangulate across data sources (interviews, documents, administrative records)
use neutral question phrasing
include time anchors (e.g., policy implementation milestones)
consider anonymous data collection approaches where appropriate

Building a sampling plan anchored in South African contexts

For South African studies, sampling plans should account for:

provincial policy variations
differing enforcement capacities
labour market segmentation (formal/informal, sectoral, rural/urban)
linguistic diversity for interviews and questionnaires

Example sampling plan (conceptual):

3 provinces: one with high compliance documentation, one with medium, one with low
2 sectors: one highly unionised, one with more fragmented labour relations
stakeholders: 6 union leaders, 6 workers, 6 employers, 6 regulators (numbers depend on your design)

The important element is that the sample enables your explanation: if enforcement differences are central to your theory, then you must sample across enforcement variation rather than only within one context.

3) Data Collection Strategies: Qualitative, Quantitative, and Mixed-Methods Design in Practice

Choosing methods based on research questions (not habits)

A frequent design weakness is selecting methods because they are familiar. SOCL7045A expects alignment between question type and method logic.

Qualitative methods fit:
- “how” and “why” questions
- meaning-making, interpretation, and process
- understanding institutional practices and decision-making
Quantitative methods fit:
- prevalence, patterns, relationships among measurable variables
- testing hypotheses or estimating effects
Mixed methods fit:
- when one method lacks sufficient evidence for the full claim
- when you need breadth plus depth
- when you need to explain unexpected quantitative patterns through qualitative inquiry

Qualitative data collection: interviews, focus groups, and documentary ethnography

Semi-structured interviews

Strength: depth and flexibility.
Design requirements:

an interview guide organised by themes aligned with your conceptual framework
pilot testing for wording and sensitivity
ethical consent processes and confidentiality planning
audio recording and transcription protocols where feasible

Interview guide structure (example template):

Background: role, experience, and context
Policy/implementation awareness: what actors know and how they learned it
Practices: what they do in specific situations
Barriers and enabling conditions
Outcomes: perceptions of impact and evidence sources
Reflections: contradictions, disagreements, and “what would change”

Focus groups

Strength: collective meaning-making and group dynamics.
Risks:

dominant voices can silence others
sensitive labour topics may heighten fear
Design solutions:
careful participant selection
neutral facilitation
split groups by employment status when retaliation concerns are relevant

Documentary and policy analysis

Strength: captures formal discourse, institutional logics, and implementation procedures.
Design requirements:

clear document selection criteria (time range, document type, relevance)
coding scheme for themes and policy mechanisms
contextual reading: how documents are interpreted and used by actors

In labour and policy research, document analysis can include:

policy frameworks and amendments
collective bargaining agreements
inspection guidelines
court judgments or arbitration records (where ethically accessible)

Quantitative data collection: surveys, administrative data, and measurement integrity

Survey design considerations

A credible survey design needs:

clear target population
questionnaire structure that minimises misunderstandings
pre-testing (cognitive interviews)
translation and back-translation if multilingual contexts are involved

Item design principles:

avoid double-barrelled questions (“and/or” pitfalls)
use consistent response formats
ensure recall time windows are explicit (e.g., “in the last 12 months”)

Administrative or secondary data

Advantages:

often large-scale and less burdensome for participants.
Risks:
missingness, definitional changes, and inconsistent reporting.
Design requirements:
data cleaning plan
variable mapping and documentation
triangulation with primary data to assess interpretation validity

Mixed-methods design: integration strategies that are not superficial

Mixed-methods failures often come from collecting qualitative and quantitative data separately without a plan to integrate.

Key integration patterns (use as design logic):

Convergent design: analyse both datasets separately, then compare/merge results.
Explanatory sequential design: quantitative results first, then qualitative work to explain patterns.
Exploratory sequential design: qualitative work first to develop measures/hypotheses, then quantitative testing.

Labour-policy case illustration:

Quantitative phase identifies that compliance is lower in certain districts.
Qualitative phase interviews reveal that compliance gaps stem from staffing shortages and unclear jurisdiction between agencies.
Integration step: incorporate qualitative mechanisms into a revised model and interpret district-level differences accordingly.

Fieldwork planning: access, trust, and risk management in South Africa

In South African labour and policy contexts, fieldwork often includes institutional entry challenges (permissions, gatekeepers) and heightened sensitivity due to fear of employment consequences.

Access and gatekeeping

Your design should include:

a pathway to gain organisational permission (e.g., via departmental approvals, union leadership, employer contacts)
an explicit plan for participant recruitment that reduces coercion risk
clarity about whether participation is voluntary and non-influence on employment decisions

Trust-building

Strategies:

spend time in the field before formal interviewing (when feasible)
clarify your role as a researcher and the boundaries of your involvement
be transparent about confidentiality limits

Risk assessment

Design a risk register that includes:

risks of retaliation
privacy risks in recorded interviews
psychological discomfort discussing workplace incidents
data storage and encryption needs

This is not administrative paperwork—risk assessment influences method choice (e.g., audio recording vs note-taking; location choice for interviews).

Data quality procedures: documentation, transcription, and audit trails

SOCL7045A-level research designs emphasise auditability: an external reviewer should see how decisions were made.

Key procedures:

maintain fieldnotes and context logs
keep a change log for sampling adjustments and instrument revisions
store raw and processed data with version control (even for small projects)
create an analytic memo trail: why coding decisions changed

Ethical design: consent as process, not form

Ethics in social research is design-embedded. It influences:

who can participate
what questions are asked
how data is stored and anonymised
how you handle disclosures

Informed consent elements

purpose and procedures
voluntary participation and withdrawal rights
confidentiality and data storage
risks and mitigations
contact details for ethics oversight structures

Confidentiality and anonymisation in labour research

Anonymisation is complex when participants are identifiable by role or location.

Practical approaches:

use role-based pseudonyms rather than job titles
generalise location (e.g., “a Gauteng district” instead of a specific municipality) when necessary
avoid quoting unique incident details that could identify participants

Managing disclosures

In contexts where participants disclose illegal practices, serious harm, or breaches of labour rights, design must specify:

what you can and cannot promise
mandatory reporting rules (if applicable to your institution/ethics clearance)
referral pathways to support services where appropriate

4) Data Analysis and Inference: From Coding to Statistical Reasoning to “Mechanisms”

Aligning analysis with the research claim

A research design must specify analysis logic before data collection ends. Analysis choices should match the claim type:

Descriptive claims → descriptive statistics, thematic summary, typologies
Explanatory claims → causal logic, pattern explanation, theory refinement
Mechanistic claims → sequencing, process tracing, chain-of-evidence logic
Interpretive-theoretical claims → deep thematic analysis and conceptual development

Qualitative analysis: coding, themes, and theory-building

Approaches to qualitative analysis

Common approaches include:

thematic analysis (code → theme → interpretation)
grounded theory (coding cycles → categories → theoretical integration)
discourse analysis (language, framing, power/ideology)
narrative analysis (story structures and meaning over time)

Your design should state which approach governs analysis and why it fits the research question.

A robust coding workflow

Familiarisation: read transcripts and develop initial impressions.
Initial coding: line-by-line or segment coding with a codebook draft.
Constant comparison: compare incidents within and across cases.
Memo writing: record analytic ideas, questions, and emerging mechanisms.
Theme development: group codes into coherent themes aligned to conceptual dimensions.
Refinement: merge/split themes; revise codebook.
Synthesis: produce an evidence-based account for the research question.

Ensuring analytic credibility

Credibility in qualitative research is strengthened by:

transparent codebook and code definitions
inter-coder reliability if working with multiple coders (where feasible)
triangulation across participant types and documents
negative case analysis (actively seeking cases that challenge themes)

Labour-policy example:
If you propose “fear of retaliation discourages complaints,” you should actively examine counterevidence:

cases where workers complain despite fear
cases where fear exists but complaints still occur through union intermediaries
This refines the mechanism into more nuanced conditional statements (e.g., fear plus access to collective representation reduces silence).

Process tracing and causal inference in qualitative work

Mechanistic explanations benefit from process logic:

identify the sequence of events
establish time ordering
test alternative explanations
use chain-of-evidence to connect observations to claims

What to document:

evidence for the start of a process
evidence for intermediate steps
evidence for the outcome
alternative pathways and why they are less supported

Quantitative analysis: models, variables, and inference pitfalls

From variables to constructs

Your quantitative analysis must preserve the link between operationalisation and inference:

outcomes must correspond to the measured indicators of your conceptual outcome
predictors should match your theory and not be arbitrary controls

Descriptive and inferential statistics

Typical steps:

Data cleaning and missing-data assessment
Descriptive statistics: distributions, cross-tabs, summary tables
Model selection: choose appropriate statistical model based on outcome type
Assumption checks: e.g., collinearity, influential observations
Interpretation: report effect direction and magnitude

Common pitfalls in social science quant work

using correlation as if it were causation
failing to account for selection bias (e.g., non-response among vulnerable workers)
constructing indexes without verifying internal consistency
ignoring design effects when sampling is clustered or weighted

Handling missing data and measurement error

Missing data can be non-random—especially when surveys on labour rights are sensitive.

Design-related best practices:

plan missing-data strategies during analysis plan
report how many values are missing for each key variable
choose appropriate methods (listwise deletion with justification, imputation, model-based handling)

Mixed-methods integration in analysis

Integration should be planned as part of analysis, not as afterthought.

Integration methods:

compare themes with quantitative patterns
use qualitative mechanisms to interpret statistical coefficients
convert qualitative categories into variables (where theoretically defensible) and test them quantitatively
build joint displays (tables/figures that align evidence streams)

A concrete integration scenario:

Quantitative model: “union density” predicts better access to grievance procedures.
Qualitative interviews: reveal union density matters because of legal literacy training and mediation networks.
Integration: revise your explanation from “union density works” to “union density proxies for access to mediation infrastructures,” supported by interview evidence.

Theory-building and refinement: moving from findings to contribution

SOCL7045A expects you to move beyond “what we found” to what it means for theory and policy debates.

A strong contribution typically:

anchors findings in conceptual and theoretical arguments
identifies conditions under which findings hold
explains contradictions and boundary cases
connects results to broader literature on labour, policy, and globalisation

5) Research Ethics, Governance, and Quality Assurance: Producing Trustworthy Knowledge in South African Institutional Contexts

Ethical governance: institutional ethics, risk, and procedural rigour

Ethical approval is not merely a bureaucratic hurdle; it is part of methodological quality. Ethics directly affects design choices and therefore the validity of findings.

Key governance components include:

ethics clearance application with detailed protocols
consent forms and participant information sheets
data management plan
risk assessment and mitigation strategies
reporting obligations for adverse events or serious disclosures

Design principle: if a method increases risk, you must either modify the method or justify why the risk is acceptable and still manageable.

Confidentiality, anonymity, and data protection for labour-policy research

Labour research often involves identifiable groups:

union leaders
employers in specific sectors
officials with identifiable job roles
workplaces with small populations

Data protection measures should be concrete:

encryption for stored files
secure server storage
restricted access controls
separation of identifiers from transcripts/data files
secure deletion processes when required by ethics protocols

Reflexivity: acknowledging researcher position without undermining evidence

Reflexivity is an analytic and ethical practice—especially when studying labour and policy where researchers may hold power relative to participants.

Reflexive dimensions include:

relationship to stakeholders (union-aligned vs employer-aligned perceptions)
social identity effects (language, class, gender)
institutional affiliations and perceived interests
how these affect recruitment and response

Design response:

reflexive memoing
training interviewers/assistants (if applicable)
piloting instruments to detect bias in question phrasing

Quality assurance: validity, reliability, trustworthiness, and transparency

A mature design articulates quality as a system.

Quantitative trustworthiness: internal and external validity

Internal validity: ensuring the evidence supports the claim within the study conditions.
- control confounds where possible
- ensure measurement accuracy
External validity: whether findings apply beyond the sample.
- justify generalisation or avoid overstating it
- report contextual constraints

Qualitative trustworthiness

Use strategies that strengthen credibility, transferability, dependability, and confirmability:

Credibility: triangulation, member checking (when appropriate), prolonged engagement (where feasible)
Transferability: thick description of context and participants
Dependability: transparent method documentation and consistent procedural practices
Confirmability: reflexive documentation and audit trails

Ethical interviewing: handling power dynamics and participant distress

Interviewing in labour settings can trigger distress or fear. Ethical design therefore includes:

interview pacing and the option to pause or skip questions
warning participants about sensitive topics
debriefing after interviews
referral links to support structures if needed

Ethical use of documentary and administrative data

Document analysis can raise ethics issues if documents include personal data or confidential information.

Design choices:

restrict to publicly available or permissioned documents
anonymise personal identifiers when quoting or paraphrasing
ensure data sharing agreements follow institutional requirements

Building an audit trail: a pathway to defensible research

A powerful study portfolio includes a clear audit trail so that claims can be traced back to evidence.

Audit trail components:

protocol and ethics approval documents
sampling rationale and sample adjustments
instrument versions (questionnaires, interview guides)
coding frameworks and coding revisions
analytic memos and model outputs (quant)
integration rationale for mixed-methods

This is the difference between “reporting findings” and “producing research.”

South African institutional context: aligning research design with local governance realities

When conducting research in South Africa, designs should account for:

institutional gatekeeping (universities, departments, public agencies, labour councils)
differences in administrative capacity across provinces
language accessibility and cultural appropriateness
labour relations complexity, including historical patterns of protest, collective bargaining, and policy contestation

Practical implication: Your consent process and participant recruitment strategy may need adaptations by setting. A design that works in one province may fail in another unless you incorporate these local governance realities into your planning.

Case study: designing an ethically robust mixed-methods labour-policy study

To consolidate key design principles, consider a hypothetical but realistic project framed in Wits labour-policy/globalisation research.

Research question:
How do mechanisms of workplace grievance handling differ between formal and informalised employment contexts, and how are these mechanisms shaped by enforcement capacity and global supply-chain pressures?

Design features:

Quantitative component: a survey comparing workers’ reported access to grievance channels and outcomes, with stratification by employment formality and sector.
Qualitative component: interviews with workers, union representatives, employers, and labour inspectors to explain observed quantitative patterns.
Document analysis: policy implementation documents and inspection guidelines to connect mechanisms to institutional logic.

Ethics and quality procedures:

risk register for retaliation and confidentiality
anonymisation protocol including location generalisation
encryption and secure storage plan
analytic memos and audit trail
integration plan: qualitative evidence explains why certain channels work or fail

Inference discipline:

the survey identifies patterns and associations
interviews provide mechanism evidence and contextual explanations
mixed-method integration avoids simplistic “qualitative confirms quantitative” claims; instead, it uses qualitative data to specify conditional mechanisms

This design illustrates how ethics, sampling, measurement, and integration must function as one system.

Consolidated Study Toolkit: How to Build and Defend a SOCL7045A-Level Research Design

1) Statement of the research claim

Write your claim as:

a question (what you seek to know)
a proposition (what you think might be true)
a contribution (why the study matters for labour, policy, and globalisation debates)

2) Map claim type to inference logic

Match:

description ↔ prevalence patterns
explanation ↔ associations and mechanisms
evaluation ↔ impact logic and comparative reasoning
interpretive theory ↔ meaning, discourse, and conceptual refinement

3) Create a concept-to-indicator matrix

For each key concept:

define it conceptually
break into dimensions
list indicators and data sources
justify why each indicator is valid for the concept

4) Sampling plan aligned to variation and evidence needs

For each stakeholder group and site:

state why it is included
state what variation it provides
state how it supports the inference logic

5) Data collection protocol with ethics embedded

Specify:

consent procedure
recording/transcription approach
participant recruitment pathway
confidentiality and anonymisation method
risk mitigations

6) Analysis plan with transparency

Specify:

analytic approach (qual, quant, mixed)
coding workflow and codebook development
quantitative models and assumption checks
integration strategy (how qualitative and quantitative results connect)

7) Quality assurance and reporting discipline

Commit to:

audit trail
reflexive documentation
triangulation rationale
limitations stated in alignment with design assumptions

Final Exam-Style Prompts (Practice for SOCL7045A)

Use these prompts to practise turning methodological concepts into defensible design arguments.

Design alignment:
Explain how your epistemological stance informs your method choice for a study on labour policy implementation in a South African province.
Sampling justification:
Propose a sampling strategy for researching grievance mechanisms across formal and informal employment, and justify why your sample supports causal/mechanistic inference.
Operationalisation:
Define “policy capacity” and propose three measurable indicators for it. For each indicator, explain what it can and cannot capture.
Ethical risk design:
Identify ethical risks in interviewing workers about workplace retaliation and propose mitigation strategies consistent with a data protection plan.
Mixed-method integration:
Design an explanatory sequential mixed-method study where survey results require qualitative follow-up. Describe the integration step and how you will avoid superficial “confirmation.”
Validity and bias:
Discuss at least two measurement biases common in labour research and how design features can reduce their impact.

These notes emphasise that social research design is ultimately about defensible reasoning: linking theory to operationalisation, methods to inference, ethics to procedure, and evidence to credible claims—within the complex labour-policy and globalisation realities of South Africa.