Psychology 771 Research Methodology Study Notes for Stellenbosch University Honours

These study notes provide a structured, exam-focused guide to Psychology 771 Research Methodology in the context of a Stellenbosch University Psychology Honours cluster. The material covers the logic of psychological inquiry, research design, measurement, ethics, data analysis, and report writing, with emphasis on the kind of conceptual precision expected in postgraduate examinations. The notes are written to support revision, answer construction, and deeper understanding of how research methodology underpins credible psychological science.

1. The Logic of Psychological Research

Psychology as a science is built on systematic observation, inference, and careful control of bias. Research methodology is not simply a list of technical procedures; it is the framework that determines whether conclusions about human thought, emotion, and behaviour are trustworthy. In a postgraduate module such as Psychology 771, the central issue is not whether a study was conducted, but whether its design allowed a valid answer to the question that was asked. This requires understanding how research questions emerge, why particular designs fit particular aims, and how epistemological assumptions shape what counts as knowledge.

1.1 What makes psychological research scientific?

Psychological research is scientific when it follows procedures that are transparent, systematic, replicable, and open to scrutiny. Scientific inquiry does not guarantee truth in an absolute sense, but it does reduce the influence of intuition, anecdote, and selective perception. In psychology, this matters because many phenomena are complex, socially embedded, and difficult to observe directly. Concepts such as intelligence, resilience, prejudice, trauma, and identity cannot be measured in the same straightforward way as temperature or mass. They must be operationalised, inferred, and interpreted.

The scientific character of research is usually reflected in several features:

Empiricism: knowledge is based on observed data rather than speculation alone.
Systematic method: procedures are planned in advance and carried out consistently.
Testability: claims can be examined against evidence.
Replication: other researchers can repeat procedures to check whether findings hold.
Falsifiability: a claim can, in principle, be shown to be wrong.
Transparency: methods, analyses, and decisions are documented clearly.

A common mistake in exam answers is to describe science only as “objective.” Objectivity is important, but psychology recognises that complete neutrality is rarely possible. Researchers bring assumptions, theoretical commitments, cultural backgrounds, and methodological preferences to the research process. For that reason, good research does not pretend bias does not exist; instead, it uses design, reflexivity, and methodological rigour to manage bias.

1.2 Research questions, aims, and hypotheses

Every project begins with a question. Good research questions are specific, relevant, feasible, and theoretically grounded. They should identify a relationship, difference, pattern, process, or meaning that can be investigated using appropriate methods. A vague question such as “What causes stress?” is too broad for a meaningful study. A stronger version would be: “How does perceived academic workload predict anxiety symptoms among first-year psychology students at a South African university?”

A research aim states the broad purpose of the study, while a research question turns that purpose into an interrogative form. A hypothesis is a testable prediction, usually derived from theory or prior evidence. In quantitative research, hypotheses are often directional or non-directional:

Directional hypothesis: predicts the direction of an effect or association.
Non-directional hypothesis: predicts that an effect or association will exist, but not its direction.

For example, a directional hypothesis might state: “Students exposed to a mindfulness intervention will report lower test anxiety than students in the control group.” A non-directional hypothesis would state only that a difference exists. The strength of a hypothesis is not in its length, but in its precision and testability.

Research questions, aims, and hypotheses must align. If the question is exploratory, a rigid hypothesis may be inappropriate. If the project is confirmatory, then a clear hypothesis is essential. In exams, showing this alignment demonstrates methodological maturity.

1.3 Deductive and inductive logic

Research methodology is shaped by two broad forms of reasoning:

Deductive reasoning moves from theory to data. A theory generates a hypothesis, and data are collected to test it.
Inductive reasoning moves from data to theory. Patterns in observations inform conceptual development.

Quantitative psychology often relies on deductive logic, especially in experimental and survey research. Qualitative research often uses inductive logic, although the distinction is not absolute. Many studies combine both. For example, a mixed-methods project may begin with a quantitative survey to identify trends, followed by interviews to explain why those trends occur.

Understanding deduction and induction is important because they influence what counts as evidence. Deductive studies ask whether data support a pre-specified claim. Inductive studies ask how meanings, experiences, or categories emerge from the data. A postgraduate student should be able to explain that the choice of reasoning affects sampling, data collection, analysis, and the role of theory.

1.4 Paradigms in psychological research

A research paradigm is a broad framework of assumptions about reality, knowledge, and method. Three influential paradigms are especially relevant in psychology:

Paradigm	Core assumption	Typical questions	Common methods
Positivist / post-positivist	Reality exists independently and can be approximated through observation	What is the relationship between variables?	Experiments, surveys, statistical modelling
Interpretivist / constructivist	Reality is socially constructed and understood through meaning	How do people interpret their experiences?	Interviews, focus groups, ethnography
Critical / transformative	Knowledge is linked to power, inequality, and social justice	Whose interests are served by dominant knowledge?	Participatory research, critical discourse analysis

The positivist tradition emphasises measurement, prediction, and causal inference. The interpretivist tradition emphasises context, subjective experience, and meaning-making. The critical tradition asks how research can expose oppression, challenge taken-for-granted assumptions, and support emancipatory change. In psychology, these paradigms are not merely philosophical labels; they determine what kind of questions are worth asking and what kind of evidence is legitimate.

A strong exam answer often shows that paradigms are not mutually exclusive in practice. A study of burnout among healthcare workers may use quantitative measures of emotional exhaustion and also qualitative interviews about workplace culture. The researcher may adopt a pragmatic stance, selecting methods according to the research problem rather than allegiance to a single philosophical school.

1.5 The importance of theory

Theory gives research structure. Without theory, studies become disconnected observations with little explanatory power. A theory is not simply a guess; it is a coherent set of concepts and propositions that explains phenomena and guides prediction. Psychological theories can be broad, such as social learning theory, or more specific, such as cognitive appraisal models of stress.

Theory matters for several reasons:

It helps refine vague questions into testable claims.
It guides the selection of variables and constructs.
It informs the interpretation of findings.
It allows findings to contribute to cumulative knowledge.

For example, if a study finds that social support predicts lower depression scores, theory helps interpret whether this is because social support buffers stress, improves coping, or alters self-appraisal. Without theory, the result remains descriptive. With theory, it becomes meaningful and connected to a wider body of knowledge.

1.6 Common exam pitfalls in this section

Students often lose marks by treating methodology as purely procedural. Examiners usually expect conceptual understanding. Common pitfalls include:

Confusing research problem with research question
Confusing aim with hypothesis
Treating induction and deduction as simple opposites rather than complementary logics
Describing paradigms without linking them to methods
Ignoring theory when discussing research design

A useful revision strategy is to practise explaining why a particular design fits a particular question. For instance, a causal question requires stronger control than a descriptive question. A question about lived experience may require depth rather than numerical generalisation. This kind of reasoning is central to research methodology.

2. Research Designs, Variables, and Sampling

Research design is the blueprint of a study. It determines how data are gathered, how variables are handled, and how confidently conclusions can be drawn. In Psychology 771, the ability to distinguish among designs is essential because exam questions often require comparison, evaluation, and justification. A design is not “better” in the abstract; it is better or worse relative to a particular question, setting, and ethical constraint.

2.1 Major types of research design

The main quantitative designs commonly discussed in psychology include experimental, quasi-experimental, correlational, cross-sectional, longitudinal, and survey designs. Each has strengths and limitations.

Experimental designs

An experimental design involves manipulation of an independent variable, control of extraneous variables, and random assignment to conditions. This design is the strongest for causal inference because it allows the researcher to isolate the effect of a variable under controlled conditions.

Example: A researcher tests whether a brief breathing exercise reduces state anxiety. Participants are randomly assigned to a breathing intervention or a control condition. Anxiety scores are measured before and after the intervention. If the treatment group improves more than the control group, the researcher has stronger grounds for causal inference.

Strengths:

High internal validity
Strong control over confounds
Supports cause-and-effect conclusions

Limitations:

May have low ecological validity
Can be expensive and time-consuming
Some variables cannot be ethically manipulated

Quasi-experimental designs

A quasi-experiment resembles an experiment but lacks random assignment. Groups may already exist, such as classes, schools, or clinical categories. This design is common in real-world psychological research where random assignment is not feasible.

Example: A university compares stress levels between students who attended a voluntary resilience workshop and those who did not. Since students self-select into groups, differences may reflect pre-existing motivation rather than the workshop itself.

Strengths:

Useful in natural settings
Often more ethically and practically feasible than experiments

Limitations:

Lower internal validity
Greater risk of selection bias and confounding

Correlational designs

Correlational research examines the relationship between variables without manipulation. It can identify whether variables co-vary, but it cannot establish causation.

Example: A study investigates the association between social media use and sleep quality among adolescents. If higher social media use is associated with poorer sleep, the study cannot determine whether social media causes sleep problems, whether sleep problems increase social media use, or whether a third variable influences both.

Strengths:

Useful for prediction and hypothesis generation
Ethical and practical for variables that cannot be manipulated

Limitations:

Cannot establish cause and effect
Vulnerable to third-variable explanations

Cross-sectional and longitudinal designs

A cross-sectional study collects data at one point in time. It is efficient and useful for describing prevalence or relationships at a specific moment. A longitudinal study collects data over time, allowing researchers to examine change, development, and temporal ordering.

Cross-sectional research is good for breadth, but longitudinal research is better for understanding trajectories. For example, a cross-sectional study may show that first-year students report more anxiety than fourth-year students, but it cannot show whether anxiety decreases as students progress through university. A longitudinal study following the same cohort over four years can address this issue.

2.2 Variables and operationalisation

A variable is any characteristic that can vary between participants, groups, or situations. Variables are central to quantitative research because they are the building blocks of measurement and analysis. Common types include:

Independent variable (IV): the presumed cause or predictor
Dependent variable (DV): the outcome or effect
Control variable: held constant or statistically adjusted
Extraneous variable: any variable that could influence the DV
Confounding variable: an extraneous variable that systematically varies with the IV and threatens interpretation

Operationalisation converts abstract constructs into measurable indicators. For example, “stress” may be operationalised through a validated scale such as a perceived stress questionnaire, physiological measures such as cortisol, or behavioural indicators such as sleep disruption. Good operationalisation requires conceptual clarity and measurement validity. A poor operational definition can undermine the entire study, even if the data collection is methodologically neat.

A useful way to think about variables is in layers:

Conceptual level: the theoretical construct
Operational level: how the construct is measured
Analytical level: how the measured variable enters statistical analysis

If these layers are not aligned, the study may answer a different question from the one intended. For example, measuring “academic performance” only through attendance records may not capture actual learning, and using a vague self-report measure may not reflect objective achievement.

2.3 Internal and external validity

Validity refers to the extent to which a study measures or demonstrates what it claims to. Two broad forms are especially important:

Internal validity: the degree to which a causal conclusion is justified
External validity: the degree to which findings generalise beyond the study context

High internal validity is achieved by controlling confounds, using random assignment, and standardising procedures. High external validity is achieved when the sample, setting, and procedures resemble real-world conditions or when the finding generalises across contexts.

These forms of validity often trade off against each other. Highly controlled laboratory studies may produce strong causal claims but limited generalisability. Field studies may be more realistic but less controlled. The key is not to assume that one form of validity automatically cancels the other. A well-designed study can maximise both to a reasonable degree, depending on the question.

2.4 Sampling principles

Sampling determines who participates and whether the sample represents the population of interest. In psychology, sampling is often constrained by convenience, access, ethics, and cost. Nonetheless, the logic of sampling must be understood carefully.

Probability sampling

In probability sampling, each member of the population has a known chance of selection. Common forms include:

Simple random sampling
Systematic sampling
Stratified sampling
Cluster sampling

Probability sampling is ideal for generalisability because it reduces selection bias and supports statistical inference. However, it is often difficult in practice.

Non-probability sampling

In non-probability sampling, selection probabilities are unknown. Common forms include:

Convenience sampling
Purposive sampling
Snowball sampling
Quota sampling

These methods are widely used in psychology because they are practical and efficient. For example, purposive sampling is suitable when the study requires participants with a specific experience, such as survivors of a particular form of trauma. However, findings from non-probability samples should be interpreted cautiously.

2.5 Sample size and power

Sample size affects the precision and reliability of estimates. Too small a sample may fail to detect real effects, while too large a sample may detect trivial differences that have little practical importance. Statistical power is the probability of detecting an effect if one truly exists. Power is influenced by sample size, effect size, significance level, and variability.

A study with low power is risky because it may produce false negatives and unstable estimates. In exam discussions, it is useful to explain that sample size is not merely a technical detail; it directly affects the credibility of conclusions. A study of anxiety reduction with 12 participants is unlikely to be as persuasive as one with 120 participants, assuming similar design quality.

2.6 Sampling errors and bias

Sampling errors arise when the sample does not accurately reflect the population due to chance, while bias refers to systematic distortion. Examples include:

Selection bias: participants differ systematically from non-participants
Attrition bias: participants who drop out differ from those who remain
Volunteer bias: people who choose to participate may be more motivated or more distressed than others
Coverage bias: some groups are excluded by the sampling frame

A strong methodological argument explains not only what bias is present, but how it affects interpretation. For example, a study on depression among university students that recruits only students who visit a counselling service may overestimate depression prevalence because the sample is self-selected and high-risk.

3. Measurement, Reliability, and Validity

Measurement is the bridge between theory and evidence. In psychology, nearly every meaningful construct must be measured indirectly, which makes measurement one of the most consequential stages of research. Poor measurement produces misleading findings regardless of how advanced the statistical analysis is. For an honours-level student, understanding measurement means recognising that numbers are only useful if they meaningfully represent the psychological phenomenon of interest.

3.1 Measurement levels

Variables can be measured at different levels:

Level of measurement	Description	Examples	Appropriate analysis implications
Nominal	Categories with no inherent order	Gender categories, diagnosis groups, language group	Frequencies, chi-square, mode
Ordinal	Ordered categories without equal intervals	Likert-type ranks, severity ratings	Median, non-parametric tests, careful interpretation
Interval	Equal intervals, no true zero	Standardised test scores, many psychological scales	Means, correlation, t-tests, ANOVA
Ratio	Equal intervals with true zero	Reaction time, number of errors, age	Full range of parametric analyses

Psychological researchers often use Likert scales and treat summed scores as approximately interval-level, especially when scales have multiple items and good psychometric properties. However, this decision should not be made casually. The appropriateness depends on scale construction, distribution, and intended analysis.

3.2 Reliability

Reliability refers to consistency. A measure is reliable if it produces stable, repeatable results under consistent conditions. Reliability does not guarantee validity, but a measure cannot be valid if it is wildly inconsistent.

Common forms of reliability include:

Internal consistency: whether items on a scale measure the same construct
Test-retest reliability: whether scores remain stable over time
Inter-rater reliability: whether different observers agree
Parallel-forms reliability: whether equivalent versions of a test yield similar scores

Internal consistency is often estimated using coefficients such as Cronbach’s alpha, although interpretation should be careful. A very high alpha does not automatically mean a scale is excellent; it may indicate redundancy among items. Likewise, a low alpha may reflect a heterogeneous construct rather than a bad scale. The meaning of reliability depends on the measure and the construct.

Why reliability matters

Suppose a depression scale is unreliable. Then two students with the same true level of depressive symptoms may receive very different scores depending on the day or the wording of the items. This introduces noise, weakens correlations, and can conceal real relationships. It also makes group comparisons unstable. Reliability therefore affects both statistical power and substantive interpretation.

3.3 Validity

Validity concerns whether a measure captures what it intends to capture. Several forms matter in psychology:

Content validity: the extent to which the measure adequately covers the construct domain
Criterion validity: the extent to which the measure correlates with a relevant external criterion
Construct validity: the extent to which the measure behaves as theory predicts
Face validity: the extent to which the measure appears appropriate on the surface

Construct validity is the most important in many psychological contexts because it concerns the theoretical meaning of the measure. A scale may look sensible, but if it does not relate to other variables in theoretically expected ways, it may not be measuring the intended construct.

A good exam response should distinguish validity from reliability. Reliability asks, “Is the measure consistent?” Validity asks, “Is the measure correct or meaningful?” A bathroom scale that always reads five kilograms too heavy is reliable but not valid. A scale that gives a different reading each time is neither reliable nor valid enough for serious use.

3.4 Common threats to measurement quality

Measurement error can arise from many sources:

Ambiguous items
Poorly translated instruments
Cultural mismatch
Social desirability bias
Acquiescence bias
Response fatigue
Order effects
Interviewer effects
Observer expectancy

In a South African context, language and cultural sensitivity are especially important. A measure developed in one cultural setting may not transfer cleanly into another without careful adaptation and validation. Meaning is not always equivalent across contexts, even when words seem similar. For example, expressions of distress may differ across linguistic communities, and a direct translation may miss local idioms of suffering.

3.5 Psychological tests and scales

Psychological instruments include achievement tests, aptitude tests, attitude scales, symptom inventories, and behavioural checklists. Their quality depends on development procedures and subsequent validation. Important steps include:

Defining the construct clearly
Generating items based on theory and literature
Reviewing items for content coverage
Piloting with a small sample
Analysing item performance
Refining the scale
Testing reliability and validity
Reassessing across contexts and populations

This process is iterative. A scale is not “finished” simply because it is published. Its usefulness depends on ongoing evidence about measurement properties. In examinations, it is often helpful to explain that psychological measures are evidence-based tools, not neutral windows into the mind.

3.6 Validity of conclusions in assessment

Research conclusions are only as strong as the data on which they rest. There are several levels at which validity can be compromised:

Construct validity: the operationalisation is poor
Statistical conclusion validity: the analysis is inappropriate or underpowered
Internal validity: a causal claim is confounded
External validity: the result does not generalise
Ecological validity: the findings do not reflect real-world functioning

A study may score well on one dimension and poorly on another. For example, a laboratory test of attention may have excellent control but limited ecological realism. A field observation may capture real behaviour but struggle with measurement precision. A sophisticated methodology answer recognises these trade-offs and argues for the best fit to the research purpose.

4. Ethics, Research Integrity, and Professional Responsibility

Ethics is not an administrative obstacle to research; it is the foundation of responsible psychological science. Psychological studies often involve vulnerable participants, intimate experiences, power differences, and potential harm. Because of this, methodology must always be paired with ethical reflection. In postgraduate work, ethical awareness signals that the researcher understands the human consequences of data collection, analysis, and publication.

4.1 Core ethical principles

The central principles of research ethics typically include:

Respect for persons: recognising autonomy and informed choice
Beneficence: maximising benefits and minimising harm
Justice: ensuring fair distribution of burdens and benefits
Fidelity and responsibility: maintaining trustworthiness and professional accountability

These principles apply throughout the research cycle, from proposal development to dissemination. Ethical conduct is not limited to obtaining approval before fieldwork. It includes the way researchers recruit participants, store data, present findings, acknowledge limitations, and deal with errors.

4.2 Informed consent

Informed consent means participants understand the nature of the study, what participation involves, possible risks and benefits, confidentiality arrangements, and their right to withdraw. For consent to be meaningful, information must be clear, voluntary, and comprehensible.

Important elements of consent include:

Purpose of the study
Procedures and duration
Risks, discomforts, or inconveniences
Expected benefits
Confidentiality and anonymity limits
Voluntary participation and right to withdraw
Contact details for queries or complaints

In psychology, informed consent can be complicated by deception studies, literacy differences, language diversity, and power relations. For example, students may feel pressured to participate if the researcher is also a lecturer. In such cases, safeguards such as independent recruitment, alternative tasks, and no-penalty withdrawal become especially important.

4.3 Deception and debriefing

Deception occurs when participants are not fully informed about the true purpose or nature of the study. It may be justified in limited cases where full disclosure would invalidate the research and the expected value is substantial. However, deception should never be used casually or to compensate for poor study design.

If deception is used, debriefing is essential. Debriefing should:

Explain the true purpose of the study
Clarify why deception was necessary
Check for participant distress
Provide support resources if needed
Offer an opportunity to withdraw data where appropriate

The ethical question is not simply whether deception “works,” but whether it is proportionate, necessary, and responsibly managed. Psychological research must not normalise manipulation of participants without serious justification.

4.4 Confidentiality, anonymity, and data protection

Confidentiality means information collected is not disclosed to unauthorised persons. Anonymity means identities cannot be linked to data at all. In many studies, true anonymity is difficult because participants may be identifiable through demographic combinations or digital records. Researchers therefore need careful data management procedures.

Key practices include:

Assigning participant codes rather than using names
Storing consent forms separately from data
Encrypting digital files
Limiting access to raw data
Removing identifying details from transcripts
Retaining data according to institutional policy

In South African research contexts, data protection is especially important when topics involve violence, HIV status, mental health, substance use, discrimination, or family conflict. Mishandling data can expose participants to harm and undermine trust in research.

4.5 Research with vulnerable populations

Ethical research with vulnerable groups requires additional caution. Vulnerability may arise from age, cognitive capacity, trauma exposure, institutional dependence, illness, economic insecurity, or social marginalisation. Vulnerability does not mean incapacity, but it does mean that consent and protection must be handled with heightened sensitivity.

Examples of vulnerable populations include:

Children and adolescents
Psychiatric inpatients
Prisoners
Refugees and asylum seekers
People experiencing acute distress
Economically disadvantaged communities

Researchers must avoid exploiting vulnerability for convenience. At the same time, excluding vulnerable populations entirely can produce gaps in knowledge. Ethical research requires balancing protection with inclusion.

4.6 Research integrity and misconduct

Ethics also includes the integrity of the scientific record. Research misconduct can take several forms:

Fabrication: making up data
Falsification: altering data, procedures, or results
Plagiarism: presenting another’s work as one’s own
Selective reporting: omitting inconvenient outcomes
p-hacking: trying multiple analyses until significance is found
HARKing: hypothesising after results are known

These practices distort evidence and damage the credibility of psychology. They are especially serious because research findings may influence interventions, policy, and public understanding. Even lesser forms of sloppiness—such as incomplete documentation, weak analytic choices, or overclaiming causal effects—can mislead readers.

4.7 Ethics in publication and authorship

After data collection, ethical obligations continue. Researchers should report methods honestly, acknowledge limitations, credit collaborators fairly, and avoid overstating implications. Authorship should reflect substantial intellectual contribution, not hierarchy or favouritism. Conflicts of interest must be disclosed where relevant.

A responsible researcher asks not only whether the study passed ethical review, but whether the publication accurately represents what happened. Honesty in reporting is itself an ethical practice.

5. Data Analysis, Interpretation, and Exam Writing

Data analysis transforms raw observations into interpretable findings. In research methodology, the goal is not to memorise statistical formulas in isolation, but to understand why particular analyses are appropriate and what conclusions they can and cannot support. A high-quality examination answer connects design, measurement, and analysis into a coherent chain of reasoning.

5.1 Descriptive and inferential statistics

Descriptive statistics summarise data. Common descriptive measures include:

Mean
Median
Mode
Range
Standard deviation
Frequency distributions

Inferential statistics allow researchers to draw conclusions from a sample about a broader population or about relationships between variables. These include t-tests, ANOVA, correlation, regression, chi-square tests, and non-parametric alternatives. The correct choice depends on the research question, the measurement level, distributional assumptions, and design structure.

Descriptive statistics are essential because they reveal patterns before any hypothesis test is conducted. A study may have a statistically significant result but be difficult to interpret if the distribution is highly skewed or if outliers dominate the pattern. Good analysis begins with understanding the data, not rushing into significance testing.

5.2 Hypothesis testing and significance

Hypothesis testing evaluates whether observed differences or relationships are unlikely to have occurred by chance under a null hypothesis. The p-value indicates the probability of observing data at least as extreme as those found, assuming the null hypothesis is true. If the p-value is below a pre-set alpha level, commonly 0.05, the result is described as statistically significant.

However, statistical significance should not be mistaken for practical importance. A tiny effect may be statistically significant in a very large sample but of little substantive relevance. Conversely, a meaningful effect may fail to reach significance in a small or noisy sample. The examiner usually rewards students who distinguish these issues.

Important related concepts include:

Type I error: rejecting a true null hypothesis
Type II error: failing to reject a false null hypothesis
Effect size: the magnitude of the observed effect
Confidence interval: a range of plausible values for a population parameter

Effect size and confidence intervals help contextualise p-values. They tell us how big the effect might be and how precise the estimate is. A methodology answer that mentions only significance without effect size is incomplete.

5.3 Correlation, regression, and prediction

Correlation assesses the strength and direction of a relationship between two variables. It does not imply causation. Regression extends this idea by examining how one or more predictors relate to an outcome and by estimating how much variance is explained.

In psychology, regression is useful for:

Identifying predictors of well-being, performance, or symptom severity
Controlling for covariates
Testing theoretical models with multiple variables
Exploring mediation and moderation

For example, a regression model might examine whether perceived social support, academic pressure, and sleep quality predict depressive symptoms among students. The interpretation must remain careful: prediction is not the same as causation. A strong predictor may still not be a cause.

5.4 Common statistical assumptions

Many parametric tests rely on assumptions such as:

Independence of observations
Normality of residuals or scores
Homogeneity of variance
Linearity
Absence of severe multicollinearity in multivariable models

Violating assumptions does not automatically invalidate a study, but it requires attention. Researchers may transform data, use robust methods, or select non-parametric alternatives. The methodological principle is simple: the analytic tool should fit the data structure, not the other way around.

5.5 Qualitative analysis and interpretation

Not all psychological research is numerical. Qualitative methods seek to understand meaning, experience, discourse, and social process. Common approaches include thematic analysis, grounded theory, interpretative phenomenological analysis, narrative analysis, and discourse analysis.

A strong qualitative study is characterised by:

Clear sampling rationale
Rich data collection
Reflexive awareness of the researcher’s role
Transparent analytic steps
Evidence of credibility, dependability, and transferability

Thematic analysis, for instance, involves familiarisation with data, generating codes, searching for themes, reviewing themes, naming themes, and producing an analytic narrative. The process is systematic but not mechanical. Interpretation requires judgement, sensitivity to context, and careful use of quotations as evidence.

Qualitative findings should not be dismissed as “just opinions.” When done rigorously, they reveal patterns of meaning that quantitative measures may miss. For example, a scale may show elevated stress among postgraduate students, but interviews may reveal that the stress is driven by supervision uncertainty, financial strain, and identity conflict—dimensions not easily captured in a survey alone.

5.6 Mixed methods integration

Mixed methods research combines quantitative and qualitative approaches in a single project. This can strengthen understanding by linking breadth with depth. The key is integration rather than mere coexistence. If the two components are isolated, the study becomes two separate projects rather than one coherent design.

Common mixed-methods structures include:

Sequential explanatory: quantitative phase followed by qualitative explanation
Sequential exploratory: qualitative phase followed by quantitative testing
Concurrent triangulation: both phases conducted in parallel and compared
Embedded design: one method nested within the other

Mixed methods are especially useful when the research problem is multifaceted, such as exploring the effects of an intervention while also understanding participants’ experiences of it. In exam writing, the value of mixed methods lies in its ability to address both causality and meaning, provided the design is logically integrated.

5.7 Writing strong exam answers

A strong Psychology 771 exam answer is usually characterised by three things: accuracy, structure, and evaluation. Accuracy means using terms correctly. Structure means organising the answer logically, often from definition to explanation to critique. Evaluation means showing insight rather than only listing facts.

A reliable answer strategy is:

Define the concept clearly.
Explain its methodological significance.
Provide a concrete psychological example.
Compare it with related concepts.
State a limitation or critical issue.
Conclude with why it matters for research quality.

For example, if asked about validity, do not stop at defining it. Explain internal, external, and construct validity; show how each can be threatened; and link the discussion to design choice and interpretation. If asked about sampling, discuss why probability methods support generalisation but are often difficult to implement, and why non-probability methods remain common in psychological research despite lower representativeness.

5.8 High-yield revision summary

The most important methodological ideas to retain are:

Research questions should determine design, not the other way around.
Measurement quality shapes the quality of all conclusions.
Reliability and validity are related but not identical.
Causation requires careful control, not just correlation.
Sampling affects generalisability and bias.
Ethics is inseparable from methodology.
Statistical significance is not the same as practical importance.
Qualitative and quantitative methods answer different kinds of questions.
Mixed methods are strongest when integrated purposefully.
Good exam answers combine definition, example, critique, and application.

A final useful habit is to practise explaining why a method is appropriate for a specific psychological problem. That kind of reasoning demonstrates mastery. A student who can compare designs, critique measurement, evaluate ethics, and interpret findings coherently is well prepared for honours-level assessment.

6. Consolidated Revision Framework for Psychology 771

Research methodology becomes manageable when the major concepts are mentally organised into a sequence. A useful way to remember the field is to think of research as a chain in which each link depends on the quality of the previous one. If the question is vague, the design will be weak. If the design is weak, the sample may be inappropriate. If the sample is poor, measurement may mislead. If measurement is poor, analysis will not rescue the study. This chain perspective is especially helpful in exam settings because it prevents isolated memorisation and encourages integrated reasoning.

6.1 The methodological chain

A robust psychological study usually follows this sequence:

Identify the problem
- A real-world or theoretical gap is recognised.
- The problem must be meaningful and researchable.
Formulate the question
- The problem is converted into a focused question.
- Key constructs are named and bounded.
Choose a paradigm
- The researcher decides whether the study is primarily quantitative, qualitative, or mixed methods.
- The philosophical stance shapes the approach.
Select a design
- Experimental, quasi-experimental, correlational, descriptive, or qualitative design is chosen.
Operationalise variables or concepts
- Abstract ideas are turned into observable indicators.
Sample appropriately
- Participants or cases are selected in line with the aims of the study.
Collect data ethically
- Consent, confidentiality, and protection are ensured.
Analyse data carefully
- Analyses match the design and the research question.
Interpret findings in context
- Results are linked back to theory, literature, and limitations.
Report responsibly

Conclusions are accurate, balanced, and transparent.

This chain is useful because it shows that methodological quality is cumulative. A study is not strong because it uses a complicated statistical test; it is strong because all stages align logically.

6.2 A practical example of methodological alignment

Consider a study examining whether peer mentoring improves adjustment among first-year psychology students at a South African university.

Problem: first-year adjustment is often difficult.
Question: Does peer mentoring improve academic and emotional adjustment?
Paradigm: primarily post-positivist if testing effectiveness quantitatively; potentially mixed methods if student experiences are also explored.
Design: quasi-experimental if students self-select into mentoring; experimental if randomly assigned.
Variables: mentoring participation is the IV; adjustment scores are the DV.
Sampling: first-year psychology students enrolled at the university.
Ethics: informed consent, confidentiality, and no academic coercion.
Analysis: compare adjustment outcomes between groups or over time.
Interpretation: improvements may reflect mentoring, but selection effects must be considered if randomisation is absent.
Reporting: state the findings with appropriate caution.

This example demonstrates why methodology is never purely technical. Every design choice affects how confidently one can talk about change, causation, and generalisation.

6.3 Common comparative distinctions to memorise

The following distinctions often appear in examinations:

Concept pair	Key distinction
Reliability vs validity	Reliability is consistency; validity is accuracy or appropriateness
Internal vs external validity	Internal validity supports causal inference; external validity supports generalisation
Experimental vs quasi-experimental	Experimental research uses random assignment; quasi-experimental research does not
Correlation vs causation	Correlation shows association; causation requires stronger evidence
Quantitative vs qualitative	Quantitative research focuses on measurement and patterns; qualitative research focuses on meaning and experience
Probability vs non-probability sampling	Probability sampling supports known selection chances; non-probability sampling does not
Descriptive vs inferential statistics	Descriptive statistics summarise data; inferential statistics draw conclusions beyond the sample

Memorising these distinctions is not enough. The best exam responses show how each distinction affects methodological strength and interpretation.

6.4 How to approach an exam question

When faced with a methodology question, a disciplined response often works best. The following approach is effective:

Identify the keyword
Determine whether the question asks for definition, comparison, critique, application, or evaluation.
Define the core concept
Use precise terminology and avoid vague language.
Develop the explanation logically
Move from general to specific, or from principle to example.
Use an example
Anchor the concept in a psychological scenario.
Evaluate or critique
Mention limitations, assumptions, and alternatives.
Conclude with methodological significance
Explain why the concept matters for validity, ethics, or interpretation.

For instance, if asked to discuss sampling, do not merely list sample types. Explain why sampling affects representativeness, how bias can arise, and how practical constraints shape real-world decisions. If asked about ethics, show how principles translate into concrete procedures.

6.5 Final conceptual anchors

Some ideas deserve to be held as core anchors because they organise much of the module:

Good research is aligned research: question, design, sample, measure, and analysis must fit together.
No method is universally best: each method is appropriate for certain purposes and limited for others.
Measurement is never neutral: every measure is a theoretical and practical choice.
Ethics is part of quality: unethical research is methodologically compromised.
Interpretation must be proportionate: results should not be claimed beyond what the design allows.
Context matters: psychology in South Africa, like psychology anywhere, must consider language, culture, inequality, access, and history.

These anchors are useful because they connect the module’s scattered technical content into a coherent intellectual frame. In postgraduate psychology, the strongest students are rarely those who merely remember definitions. They are the ones who can explain how methodological decisions produce knowledge, where those decisions can fail, and why rigorous research remains essential to psychology as a discipline.

6.6 Last-minute revision checklist

Before an assessment, it helps to review whether you can answer the following:

Can you distinguish a research problem from a research question?
Can you explain when an experimental design is preferable to a correlational one?
Can you define and compare reliability and validity?
Can you identify common sampling biases?
Can you explain informed consent, deception, and debriefing?
Can you distinguish statistical significance from practical significance?
Can you describe how qualitative and mixed methods differ from quantitative research?
Can you critique a study using methodological language rather than common sense alone?

If the answer to these questions is yes, the methodological foundations are strong. If not, the weak points should be revised with examples, not just definitions. Methodology is learned by applying concepts to cases, comparing alternatives, and repeatedly asking whether a design truly answers the question it claims to answer.