Wits SOCL2005A Social Research Design & Practice: Quantitative & Qualitative Notes

SOCL2005A—Social Research Design & Practice: Quantitative & Qualitative Notes—is a course focused on how sociologists build rigorous research designs, choose appropriate methods, and justify methodological decisions. At Wits Sociology, the course typically expects you to demonstrate competence in both quantitative and qualitative approaches, and to show how each method answers different kinds of research questions. These exam notes consolidate core concepts, design logic, and practical research workflows used in social science, with emphasis on the South African higher education environment (universities and TVETs) where fieldwork access, ethics, and sampling realities shape what “good research” looks like.

1) SOCL2005A Research Design Fundamentals (Quant–Qual as a Single Toolkit)

Research design is not just a technical plan; it is the intellectual bridge between (1) what you want to know, (2) why it matters sociologically, and (3) how you will produce evidence credible enough for critical evaluation. In a mixed-world like South Africa—where inequality, language variation, institutional constraints, and uneven access to data are everyday realities—design quality often depends on anticipating “failure modes” (non-response, measurement error, gatekeeper bias, ethical risks) before they occur.

1.1 The logic chain: Question → Conceptualisation → Evidence → Inference

A strong SOCL2005A answer usually shows that you can articulate the full chain:

1. Research question (RQ)

   • “What is happening?” (descriptive)
   • “Why does it happen?” (explanatory)
   • “How do people experience it?” (interpretive)
   • “How do institutions shape it?” (structural + process)

2. Conceptualisation

   • Define the constructs (e.g., “student belonging,” “food insecurity,” “institutional support”).
   • Decide whether constructs are measured (quant) or understood (qual), or both.

3. Operationalisation / indicators

   • Quant: identify measurable indicators (scales, indices, variables).
   • Qual: identify observable practices, meanings, and contexts, plus how you will record them.

4. Evidence generation

   • Quant: survey, structured observations, administrative data.
   • Qual: interviews, focus groups, ethnography, document analysis.

5. Inference

   • How you will move from data to claims:
     • Quant: inference through statistical relationships, estimation uncertainty, and assumptions.
     • Qual: inference through interpretation, plausibility, and analytic transparency.

A common exam pitfall is listing methods without demonstrating the inference logic. Markers often reward students who can connect design choices to credibility.

1.2 What “good design” means in social research

Good design is judged across multiple dimensions:

• Validity: Are you measuring/understanding what you claim to measure/understand?
• Reliability: Would you get similar findings under similar conditions?
• Ethics: Are you protecting participants and ensuring consent, confidentiality, and minimisation of harm?
• Feasibility: Can you actually conduct the design with realistic time, access, and resources?
• Transparency and replicability: Can another researcher follow your logic and decisions?

In South Africa, feasibility and ethics are not “soft constraints”—they directly affect the design. For example, if gatekeepers at a TVET restrict access or require specific scheduling, the sampling and fieldwork timeline may need adjustment. Similarly, when sensitive topics involve violence, undocumented status, or mental health, confidentiality cannot be treated as a generic checkbox; it must be operationalised in how data are stored and anonymised.

1.3 Typical SOCL2005A design frameworks you may be expected to use

You should recognise and explain several standard design frameworks:

• Cross-sectional vs longitudinal designs

  • Cross-sectional: collect data at one point (e.g., survey of 2nd-year students in 2026).
  • Longitudinal: repeat data collection (e.g., student transitions across years).

• Experimental vs quasi-experimental logic

  • Randomised control trials are rare in routine social research but “natural experiments” and quasi-experimental methods may be discussed in exam contexts.
  • Quasi-experimental designs use comparison groups without full randomisation.

• Descriptive vs explanatory vs exploratory designs

  • Exploratory: build understanding first (often qualitative) to refine variables and hypotheses later.

• Mixed-methods designs

  • Triangulation: use multiple methods to assess the same phenomenon.
  • Complementarity: one method strengthens understanding of the other.
  • Expansion: each method adds unique insights.

A high-scoring approach in exams is not “mixed methods is good”; instead it is mixed methods is justified by the research question and the limitations of each method.

1.4 Building a research proposal: the examiner’s checklist

Many SOCL2005A exam tasks (or tutorial expectations) reward a consistent proposal structure. A strong structure includes:

• Title and RQ
• Background (why it matters in SA context)
• Key concepts and definitions
• Literature-informed rationale (what is known; what is missing)
• Hypotheses or guiding research questions
• Methodology:
  • Sampling plan
  • Data collection tools
  • Fieldwork plan and timeline
  • Data analysis plan
• Ethics and risk management
• Validity/reliability strategies and limitations
• Expected contributions

You do not need to use exactly these headings, but your answer should “cover the same ground.”

2) Quantitative Social Research: Sampling, Measurement, Surveys, and Causality Limits

Quantitative research in SOCL2005A focuses on how to design surveys and analyses that allow sociological claims to be evaluated. You will typically be expected to handle sampling, measurement, questionnaire design, descriptive statistics, inference, and some understanding of causality and its limits in observational research.

2.1 Research questions suited to quantitative designs

Quantitative methods are most appropriate when you want:

• Patterns, distributions, and comparisons
  Example RQ: “How does perceived institutional support differ across faculties at Wits?”
• Relationships among variables
  Example RQ: “Is food insecurity associated with academic performance?”
• Testing hypotheses
  Example hypothesis: “Students who report higher social support have higher likelihood of persisting to the second semester.”

Quantitative designs can also support explanatory claims, but you must acknowledge assumptions and alternative explanations.

2.2 Sampling strategies: from theory to field reality

2.2.1 Probability sampling vs non-probability sampling

• Probability sampling (each unit has a known chance)

  • Benefits: supports statistical inference to a population.
  • Common forms:
    • Simple random sampling
    • Stratified sampling
    • Cluster sampling
    • Systematic sampling

• Non-probability sampling

  • Benefits: often feasible when access is limited.
  • Common forms:
    • Convenience sampling
    • Purposive sampling
    • Quota sampling

In South African universities and TVETs, probability sampling may be constrained by timetables, access policies, and the availability of sampling frames (e.g., enrolment lists). Exams may ask you to propose what you can do instead while still being methodologically responsible.

2.2.2 Example: Designing a student survey sample at a South African campus

Suppose a researcher wants to study first-year students’ experiences of academic advising at a large university campus. A probability approach could be:

• Build a sampling frame from enrolment lists (where permissible).
• Stratify by:
  • Faculty (e.g., Humanities vs Science)
  • Gender category (if recorded consistently)
  • Residence status (residence vs off-campus)

Then within each stratum, randomly select students to receive a survey.

If probability sampling is impossible, a feasible non-probability approach could be:

• Use purposive sampling to select classes in each faculty.
• Administer questionnaires to students during class sessions.
• Document the limitations: students absent from those sessions may differ systematically (attendance bias).

Exam tip: A strong answer explicitly states what bias you might be introducing and how you would minimise it (e.g., reschedule sessions, include multiple time slots, compare sample demographics to known population demographics).

2.3 Questionnaire design and measurement validity

2.3.1 From constructs to indicators

Consider a construct like “sense of belonging”. You must define what belonging means:

• Emotional attachment to the institution?
• Perceived support from staff and peers?
• Feeling recognized and safe?

Then create indicators:

• Likert items: “I feel connected to students at this institution.”
• Behavioural proxies: participation in student committees
• Experience-based items: “Academic staff treat me with respect.”

Measurement validity depends on whether the items actually capture the construct.

2.3.2 Types of measurement

• Nominal variables (gender category, field of study)
• Ordinal variables (Likert scale responses, ranks)
• Interval/ratio variables (income amounts, GPA/marks, attendance counts)

A common exam requirement is to know what kinds of statistical operations are appropriate for each level.

2.3.3 Scale construction: reliability and internal consistency

When using multi-item scales, you should recognise reliability concepts:

• Internal consistency (often assessed using Cronbach’s alpha in many introductory courses).
• Item-total correlations (items that do not align reduce reliability).
• Reverse-coded items (ensure correct interpretation; mis-coded reverse items can distort analysis).

If you are asked in exams how to improve reliability, answer options include:

• Use well-established validated scales (if available).
• Pilot test the questionnaire.
• Revise ambiguous wording.
• Ensure translation accuracy across languages.

2.4 Questionnaire wording, bias, and respondent cognition

2.4.1 Common sources of measurement error

• Social desirability bias
  Respondents may report “acceptable” answers rather than truthful experiences (e.g., claiming regular attendance).
• Recall bias
  Questions requiring memory over long periods can reduce accuracy.
• Order effects
  The sequence of items can influence responses.
• Leading questions
  Embedded assumptions can bias responses.

2.4.2 Designing Likert items: practical rules

Likert items should be:

• Consistent in direction (avoid mixing strongly agree/strongly disagree meanings without careful coding).
• Balanced in response options.
• Clear about time frames (“in the last month” vs “in general”).

In South African multilingual contexts, clarity is essential. If your survey is available in multiple languages, the translation must preserve meaning rather than literal words.

2.5 Descriptive statistics as sociological storytelling

Quantitative research is not only hypothesis testing. Descriptive statistics allow sociologists to map patterns of inequality and experience.

Common descriptive outputs:

• Frequencies and percentages (e.g., % reporting high advising satisfaction)
• Measures of central tendency (means/medians)
• Measures of dispersion (standard deviation, interquartile range)
• Cross-tabulations (e.g., satisfaction by residence status)
• Visualisations (bar charts, histograms, box plots)

In exam responses, markers like when you interpret. For example:

• Not just “mean belonging score is 3.2,” but “belonging scores cluster around moderate agreement, suggesting many students experience partial connection rather than strong attachment.”

2.6 Inferential statistics: what you’re testing and why

Introductory exam contexts may include logic about:

• Correlation (association, not causation)
• Differences between groups (t-test/ANOVA logic)
• Regression (model-based explanation of relationships)

Even if you do not need to compute numbers, you should explain:

• What a test statistic tells you
• What a p-value implies (and what it does not)
• Confidence intervals as uncertainty ranges for estimated effects

2.6.1 Causality limits in observational quant research

A major quantitative exam theme is: correlation is not causation. You must show understanding of:

• Confounding variables (third factors influence both X and Y)
• Selection bias (the sample is not comparable)
• Reverse causality (Y influences X)
• Measurement error (weak measurement attenuates relationships)

In sociological research, these issues are especially important because social phenomena are multi-causal.

2.7 Worked example (conceptual): operationalising and analysing an inequality topic

Imagine a researcher studying relationship between household food insecurity and academic performance among TVET students.

• Construct: Food insecurity

  • Indicators: difficulty affording meals, skipping meals, relying on social grants.
  • Could be measured via an index (sum of items).

• Outcome: Academic performance

  • Could be self-reported marks or actual module results (if accessible).
  • If using self-report, measurement bias is possible.

• Potential confounders:

  • Employment status
  • Distance to campus
  • Access to NSFAS/bursaries
  • Study time

A responsible quantitative design would:

1. Pilot test food insecurity items to ensure comprehension.
2. Use appropriate variable coding.
3. Analyse:
   • Descriptive: compare food insecurity prevalence across campuses/programmes.
   • Regression: estimate association with academic performance while controlling for confounders.

In exams, you should also mention limitations: without random assignment, you cannot fully rule out unobserved confounding, but you can strengthen credibility through careful measurement and model specification.

3) Qualitative Social Research: Sampling for Meaning, Interviewing, Thematic Analysis, and Trustworthiness

Qualitative research in SOCL2005A aims to understand meanings, practices, and social processes in context. Where quantitative work tends to seek patterns across many cases, qualitative work often focuses on depth and explanation—how and why participants make sense of their experiences.

3.1 When qualitative methods are the best fit

Qualitative designs are appropriate when:

• You need to explore how people interpret events
• You want to understand institutional processes from inside participants’ accounts
• You expect complexity and variability not captured by pre-set variables
• The topic is sensitive and you need to establish relational trust
• There is limited prior research locally, requiring exploratory understanding

Examples suited to qualitative studies in South African education and training contexts:

• How students describe barriers to learning when funding delays occur
• How lecturers explain curriculum changes and how those changes affect classroom interactions
• How TVET learners understand employability and career pathways

3.2 Sampling in qualitative research: purposive logic, information richness, and practical access

Unlike quantitative probability sampling, qualitative sampling aims for information richness.

3.2.1 Purposive sampling strategies

• Maximum variation sampling: select diverse cases to capture different experiences.
• Typical case sampling: focus on what “most” participants look like.
• Critical case sampling: select cases where something important is likely to occur.
• Snowball sampling: participants refer others (common in hard-to-reach populations).

3.2.2 Sample size and saturation (and what to say in exams)

Exams often ask about “how many interviews?” The best answer is not a fixed number; it depends on:

• Complexity of the topic
• Heterogeneity of participants
• Quality and depth of interviews
• Whether you reach thematic saturation (no meaningful new themes emerge)

A robust exam answer might say:

• Start with purposive recruitment across key categories (e.g., gender, programme type, year of study).
• Conduct interviews iteratively.
• Stop when additional interviews do not add substantially new insights.

You should avoid implying saturation is automatic. It’s an analytic judgment supported by ongoing note-taking and coding.

3.3 Interviewing as a method: craft, ethics, and power

3.3.1 Structured, semi-structured, and unstructured interviews

• Structured interviews: fixed questions, often similar to surveys.
• Semi-structured interviews: core questions plus flexibility to probe.
• Unstructured interviews: broader conversational focus, high flexibility.

SOCL2005A typically expects familiarity with semi-structured interviewing because it balances comparability and depth.

3.3.2 Question design for qualitative interviews

Effective interview questions should be:

• Open-ended (“Tell me about…”)
• Specific yet non-leading (ask about experiences and contexts without suggesting answers)
• Time-aware (“When did it start?” “What changed?”)
• Probe-capable (“How did that make you feel?” “What happened next?”)

A strong probe set includes:

• Clarification probes: “What do you mean by…?”
• Elaboration probes: “Can you give an example?”
• Contrast probes: “Was it different in the past?”
• Process probes: “How did it unfold?”

3.3.3 Managing power, language, and rapport

In South African education settings, power dynamics arise from:

• Student vs lecturer relationships
• Researcher vs participant status
• Gatekeeper influence (institution officials)
• Language barriers and translation

Ethical interviewing practices include:

• Transparent consent and voluntary participation
• Allowing participants to skip questions
• Ensuring confidentiality (including in recordings and transcripts)
• Using language accommodations (translation or bilingual interviewing where needed)

3.4 Focus groups: benefits and risks

Focus groups can generate:

• Collective norms (how groups talk)
• Interactional insights (how participants build shared narratives)
• Efficient data collection across multiple participants

Risks include:

• Dominant participants suppress others
• Participants may feel pressured to conform
• Confidentiality is harder due to participant-to-participant knowledge

Mitigation strategies:

• Skilled moderation
• Ground rules about respectful participation
• Careful recruitment to reduce power imbalances within the group
• Debriefing and guidance on confidentiality limits

3.5 Observational and document-based qualitative approaches

While interviews are common, SOCL2005A may also discuss:

• Participant observation / ethnography: understand lived experience in situ.
• Non-participant observation: record interactions without participation.
• Document analysis: policies, student handbooks, meeting minutes, institutional reports.

Document analysis requires critical reading:

• Who produced the document?
• For what purpose?
• What is absent (silences)?
• How does the document compare with participant experiences?

3.6 Qualitative data analysis: coding, themes, and analytic integrity

3.6.1 From transcript to codebook

A typical qualitative workflow:

1. Transcription and familiarisation
2. Initial coding
   • Line-by-line or concept-based
3. Focused coding and categorisation
4. Theme development
5. Interpretation and evidence selection

A codebook should be:

• Consistent (code definitions are clear)
• Revisable (codes evolve with new insights)
• Traceable (you can link themes back to coded excerpts)

3.6.2 Thematic analysis: “what” vs “how” frameworks

Thematic analysis can operate at different interpretive levels:

• Semantic themes: explicit meanings stated by participants.
• Latent themes: underlying assumptions, power relations, ideologies.

Exams may expect you to differentiate between simply summarising content and doing analysis of meaning.

3.6.3 Ensuring trustworthiness (credibility, transferability, dependability, confirmability)

A strong qualitative answer includes trustworthiness criteria:

• Credibility: confidence in truth value of findings
  • Use member checking (where appropriate), triangulation, prolonged engagement.
• Transferability: applicability to other contexts
  • Provide thick description so readers can judge similarity.
• Dependability: stability of findings over time
  • Maintain audit trails of coding decisions.
• Confirmability: degree to which findings reflect participants rather than researcher bias
  • Reflexivity journals and analytic memos.

3.7 Reflexivity: why it matters in Wits-style qualitative research

Reflexivity is not “self-disclosure for its own sake.” It is systematic attention to:

• Your positionality (e.g., language background, social class, institutional affiliation)
• How your assumptions shape questions and interpretation
• How participants might respond differently due to perceived researcher identity

In South African settings, reflexivity often includes sensitivity to:

• Historical context (e.g., apartheid legacies affecting institutional trust)
• Current resource inequalities
• Participant vulnerability and research fatigue

3.8 Example: Analysing interviews about academic advising barriers

Imagine interviewing TVET learners about difficulty accessing academic advice.

Potential themes might include:

• Visibility and gatekeeping: learners do not know where to go.
• Relational distance: interactions feel intimidating or bureaucratic.
• Language and comprehension: support materials are not accessible.
• Time and schedule mismatch: advising happens during times learners cannot attend.
• Hope and coping strategies: learners develop peer-support networks.

A strong exam response does not list themes; it explains:

• How themes are derived from coded excerpts
• How themes connect to broader sociological mechanisms (e.g., institutional structures)
• Alternative interpretations (e.g., advising is available but learners may not perceive it as relevant)

4) Mixed Methods and Applied Design Practice in South African Education Contexts

SOCL2005A often expects students to demonstrate not only separate quant/qual skills, but also the ability to integrate them into coherent research designs. Mixed methods is frequently framed as combining strengths: quantitative breadth and qualitative depth. However, integration must be intentional; otherwise, you end up with two disconnected studies rather than a single design.

4.1 Mixed methods purposes: triangulation, complementarity, expansion, and explanation

Use mixed methods when it helps answer questions that cannot be answered well by a single approach.

4.1.1 Triangulation (converging evidence)

Example: measure student belonging via a scale and also interview students about how belonging is experienced.

• Quant finds “moderate belonging scores.”
• Qual explains the reasons: recognition from lecturers, peer support, and institutional communication.

Triangulation can strengthen credibility when the two methods align; if they diverge, divergence becomes analytically interesting.

4.1.2 Complementarity (different aspects)

Example:

• Quant measures prevalence: % of students reporting meal insecurity.
• Qual explores meaning: how students interpret the impact on study motivation and stress.

Here, quant gives scale and qual explains mechanisms.

4.1.3 Expansion (sequential breadth + depth)

Example:

1. Start with survey results to identify patterns (e.g., food insecurity predicts lower attendance).
2. Conduct follow-up interviews to understand the lived processes linking food insecurity to attendance.

4.1.4 Explanation (quant-informed qual or qual-informed quant)

Explanation designs can be:

• Quant → Qual: choose variables based on statistical patterns, then explore how participants interpret them.
• Qual → Quant: identify themes from interviews and convert into survey items.

4.2 Designing a coherent mixed-methods project: integration points

A top exam answer should specify integration at:

• The sampling stage
  Do you sample the same participants for both strands or different groups?
• The data collection stage
  Are data collected simultaneously or sequentially?
• The analysis stage
  Do you merge datasets or use side-by-side comparison?
• The interpretation stage
  How will you reconcile conflicting findings?

4.3 Example case study: Investigating barriers to learning in a TVET college

A mixed-methods study might investigate barriers to learning among TVET learners experiencing chronic attendance challenges.

4.3.1 Quantitative strand

• Outcome: self-reported attendance consistency (e.g., number of classes missed in a month).
• Predictors: transport cost burden, household responsibilities, perception of lecturer support.
• Sampling: purposive selection of programmes in the college, with quota sampling by year.

Analysis:

• Descriptive statistics show high prevalence of missed classes.
• Regression estimates which factors predict attendance.

4.3.2 Qualitative strand

Conduct semi-structured interviews with a subset of learners across programme types and attendance patterns:

• High attendance
• Low attendance

Explore:

• How learners explain missed classes
• How they access support (if at all)
• How institutional communication affects decisions

4.3.3 Integration

• If transport cost is statistically associated with lower attendance, qual explores the lived decision-making:
  • Which costs are most burdensome?
  • How do learners choose between work and study?
  • How do they negotiate with family expectations?

This integration produces a more complete sociological account: not only “transport matters,” but “transport shapes choices through constraints and institutional responses.”

4.4 Triangulation vs consistency: handling divergence responsibly

A high-scoring exam response addresses what to do if methods diverge:

• Perhaps the quant instrument misses nuances captured in interviews.
• Perhaps interviews occur in a different time context (e.g., after a policy change).
• Perhaps participants underreport in survey due to social desirability or fatigue.

Rather than forcing alignment, explain divergence as part of sociological reality and potential methodological issues. Markers often reward critical realism.

4.5 Ethical integration in mixed methods: confidentiality across datasets

Ethics must be handled carefully when combining data types:

• Interview recordings can identify participants even after anonymisation.
• Survey responses may link to identifiable information if not properly de-identified.
• In small programmes or small cohorts, anonymity is harder.

Practical ethical steps:

• Separate datasets with different coding schemes.
• Remove direct identifiers early.
• Store data securely with controlled access.
• Use aggregation for reporting (especially in qualitative excerpts tied to small groups).

4.6 Researcher safety and fieldwork logistics in South African settings

Applied practice in education research includes safety considerations:

• Fieldwork timing around semester schedules and exams
• Travel risk for remote sites
• Managing contact with gatekeepers
• Ensuring you do not coerce participation through institutional authority

In exams, you may be asked “what risks exist?” Your answer should distinguish:

• Risks to participants (emotional distress, privacy breaches)
• Risks to researcher (safety, exposure to conflict areas)
• Risks to research integrity (access denial, data loss, low response rates)

5) Exam-Style Skills: Hypotheses, Validity, Reliability, and Presenting Findings (with Quant–Qual Rigor)

This final section consolidates exam-relevant skills: how to formulate hypotheses, how to evaluate validity and reliability, how to manage limitations without undermining your work, and how to present results coherently. It also emphasises “what markers reward” in SOCL2005A-style assessments: clarity, justification, and methodological literacy.

5.1 Hypotheses and guiding questions: matching form to method

5.1.1 Quantitative hypotheses

A typical quantitative hypothesis structure:

• Directional vs non-directional hypotheses
• Variable definitions
• Expected relationship

Example (conceptual):

• H1: Higher perceived academic support is associated with higher student persistence.
• H0: There is no relationship between perceived academic support and persistence.

Even if you do not compute results, you should show you understand hypothesis logic.

5.1.2 Qualitative research questions (not hypotheses)

Qualitative work uses guiding questions such as:

• “How do students describe the role of academic support in their decision to persist?”
• “What meanings do learners attach to advice from lecturers and administrators?”

The key distinction: qualitative questions explore processes and meanings rather than test predetermined statistical claims.

5.2 Validity and reliability: distinguishing them properly

5.2.1 Validity

Validity addresses whether conclusions are credible.

• Content validity: items cover all aspects of a construct.
• Construct validity: the construct is conceptually represented.
• Criterion validity: the measure aligns with external criteria (if available).
• Internal validity: causal claims are credible within the design (in quant research).
• External validity: generalisation beyond the sample.

In qualitative work, analogs include credibility and transferability rather than statistical generalisation.

5.2.2 Reliability

Reliability addresses consistency.

• In quant: stability of measurement, internal consistency of scales, standardised administration.
• In qual: dependability and consistency of coding, often supported by an audit trail and codebook discipline.

In exams, a common mistake is treating reliability as automatically guaranteeing validity. You must emphasise that consistent measurement can still measure the wrong construct.

5.3 Bias, confounding, and threats to trustworthiness: what to say under pressure

A high-quality SOCL2005A exam answer lists threats and mitigations.

5.3.1 Quantitative threats and mitigations

• Sampling bias
  Mitigation: improve recruitment coverage; use comparison checks to known demographics where possible.
• Non-response bias
  Mitigation: track response rates; compare respondents to non-respondents (if data exists); adjust recruitment.
• Measurement error
  Mitigation: pilot test; use validated items; refine translations; ensure clear question wording.
• Confounding
  Mitigation: control for confounders in regression; carefully justify included variables; consider sensitivity analyses where appropriate.
• Common method bias (if all variables come from same survey)
  Mitigation: vary response formats or time frames; use objective measures where possible.

5.3.2 Qualitative threats and mitigations

• Interviewer bias
  Mitigation: neutral probing; training; reflexive practice.
• Social desirability in interviews
  Mitigation: build rapport; assure confidentiality; use indirect questioning.
• Gatekeeper bias
  Mitigation: negotiate access transparently; recruit independently when allowed; document gatekeeper influence.
• Researcher interpretive bias
  Mitigation: audit trail, memoing, coding checks, triangulation, peer debriefing.

5.4 Presenting quantitative findings: what counts as good reporting

A coherent results presentation should include:

• Sample description (demographics, response rates if relevant)
• Key descriptive statistics
• Model outputs (direction and significance where relevant)
• Interpretations tied back to RQ and constructs
• Uncertainty and limitations

Even without exact numeric values, an exam response can demonstrate competence by:

• Explaining what a coefficient means in plain language
• Linking the model to the hypothesis
• Stating what assumptions are required (e.g., linearity, absence of severe multicollinearity, proper coding)

5.5 Presenting qualitative findings: evidence-based narrative

Good qualitative reporting includes:

• Clear theme structure
• Quotations or paraphrased evidence
• Analytical explanation: “what does this theme mean sociologically?”
• Consideration of counter-examples or alternative cases

A common mistake is overloading quotes without analysis. Another mistake is summarising without anchoring claims in data. Balance is crucial.

5.6 Combining results in a mixed-methods report: integration writing

When presenting mixed-methods findings, you should avoid:

• Reporting quant results and qual results as separate chapters with no connection
• Restating the same conclusions repeatedly

Instead, use integration writing such as:

• “The survey shows X. Interviews clarify how participants experience X by describing Y mechanisms.”
• “Where survey and interviews diverge, plausible explanations are…”

5.7 Limitations: how to write them credibly (and not defensively)

Exams often include “evaluate the study design” or “discuss limitations.” A good limitations section:

• Names the limitation clearly
• Explains why it matters (what bias it may introduce)
• Offers mitigations or ways the limitation could be addressed in future research

Examples of credible limitations:

• “The study uses non-probability sampling, which limits statistical generalisation.”
• “Self-reported marks may be affected by recall and social desirability.”
• “Interview data reflect participants’ accounts at the time of interview and may shift with policy changes.”

The tone should be scholarly, not apologetic.

5.8 Time management and workflow planning (a frequent hidden exam criterion)

Even if your exam answer focuses on methods, markers often reward realistic workflows:

1. Finalise RQ and constructs
2. Conduct a short pilot (questionnaire testing or interview guide testing)
3. Recruit participants (document recruitment process)
4. Collect data (maintain consistent procedures)
5. Analyse (coding discipline; statistical model logic)
6. Integrate findings (link mechanisms and patterns)
7. Write clearly with evidence and citations

In South African education contexts, scheduling around academic calendars is critical. If your fieldwork plan ignores semester timing, you risk infeasibility—an issue that evaluators may notice.

5.9 Mini “exam answers” bank: ready-to-adapt phrases and structures

The following are adaptable structures that help produce well-graded responses:

5.9.1 Sampling justification template

• “A [probability/non-probability] sampling approach is appropriate because…”
• “This approach allows [credibility/generalisation/feasibility] given constraints at…”
• “Potential biases include… and mitigation strategies include…”

5.9.2 Validity and reliability template

• “To ensure validity, I would… (content/construct/criterion arguments).”
• “To ensure reliability, I would… (pilot testing, consistent administration, coding discipline).”
• “Despite this, limitations remain because… (confounding/transferability etc.).”

5.9.3 Mixed methods integration template

• “Quantitative results indicate…”
• “Qualitative findings explain…”
• “Where findings diverge, this suggests either… or…”
• “Together, the design provides a more complete account of…”

5.10 Institutional context in South Africa: why design choices differ across universities, colleges, and TVETs

SOCL2005A expects you to show awareness that research settings shape design. In South Africa:

• Data access may differ (administrative records availability, institutional permission processes).
• Language diversity affects questionnaire comprehension and consent processes.
• Ethical oversight may require careful documentation and institutional engagement.
• Student and learner mobility may affect longitudinal feasibility.
• Resource constraints influence recruitment schedules and response rates.

Your exam answers should not treat “context” as generic; it should appear as concrete design implications.

5.10.1 Example: institutional permission as a design variable

If an institution delays ethics approval, your data collection may shift into a period when students are preparing for exams. That can lead to:

• Lower availability for interviews
• Increased dropout for surveys
• Changes in participants’ stress levels (confounding emotional state with perceptions)

A top exam response acknowledges that time is part of design, not an administrative detail.

Final consolidation (what to remember for exams)

• Design starts with the question: choose methods based on what the question requires.
• Quantitative rigor: sampling choices, measurement validity, and cautious inference.
• Qualitative rigor: purposive sampling, ethical interviewing, disciplined coding, and trustworthiness.
• Mixed methods integration: purposeful combination with clear points of integration and honest handling of divergence.
• South African context matters: access, language, ethics, and institutional calendars shape feasibility and credibility.

These notes support constructing exam-ready answers that demonstrate both methodological competence and sociological reasoning—exactly the combination SOCL2005A is designed to assess.