Quantitative research answers questions about how many, how much and how strongly things are related, using numbers collected in a structured way. In South African honours and master’s dissertations it usually means a survey with a validated questionnaire, analysed in SPSS, R or jamovi, and it is the approach most examiners judge most mechanically: is the design right for the question, is the sample defensible, are the instruments valid and reliable, and do the statistical tests match the data? This guide walks through each of those decisions in the order you will make them, with the sentences your methodology chapter needs. If you are setting up the chapter now, Tesify helps you structure your dissertation chapter by chapter.
What quantitative research is
Quantitative research measures variables, the characteristics that differ between people, organisations or places, and uses statistics to describe them, compare groups or test relationships. It is built on clear definitions decided in advance: what exactly you will measure, with which instrument, in which sample, and with which test. That pre-specification is its strength. It allows another researcher to repeat your study and lets readers judge how far your findings generalise.
Use quantitative research when your question includes words such as “the relationship between”, “the effect of”, “the extent to which”, “the level of” or “differences between groups”. If your question asks how people experience or make sense of something, a qualitative design is the better fit.
Step 1: Turn your question into variables and hypotheses
Every quantitative study rests on variables you can measure. A study asking whether work engagement predicts intention to leave among nurses in Gauteng public hospitals (illustrative) has:
- an independent variable (work engagement), the presumed cause or predictor;
- a dependent variable (turnover intention), the outcome;
- possible control variables (age, years of experience, unit type) that might also affect the outcome.
From these come hypotheses: H1: work engagement is negatively related to turnover intention. Each hypothesis must be testable with a specific statistical test on the data you plan to collect. A simple matrix with one row per hypothesis, listing the variables, instrument items and planned test, lets your supervisor check the logic at a glance.
Step 2: Choose a quantitative research design
| Design | What it does | Typical dissertation use |
|---|---|---|
| Descriptive (cross-sectional survey) | Describes levels and patterns at one point in time | Prevalence of burnout among teachers in one district |
| Correlational | Tests relationships between variables without manipulating them | Engagement and turnover intention among nurses |
| Comparative (causal-comparative) | Compares existing groups on an outcome | Financial literacy of students with and without bursaries |
| Quasi-experimental | Compares groups before and after an intervention without random assignment | A study-skills workshop in one class compared with another |
| Experimental | Randomly assigns participants to conditions | Rare in coursework master’s because of time and ethics constraints |
Most South African master’s research reports use a cross-sectional survey with a correlational or comparative design. Be careful with causal language: a correlational design shows association, not cause. Write “was associated with” or “predicted”, not “caused”.

Step 3: Define the population and sample
Name your target population precisely (permanent professional nurses in three named public hospitals, not “nurses in South Africa”) and state its size if known. Then choose a sampling technique:
- Probability sampling (simple random, stratified, cluster) gives every member a known chance of selection and supports generalisation. It needs a sampling frame such as a staff list.
- Non-probability sampling (convenience, purposive, snowball) is common when no list exists or access is limited. It is acceptable in many dissertations, but you must acknowledge that results may not generalise to the population.
Sample size should be justified, not guessed. Common approaches are a power analysis for your main test (for example with G*Power), a published sample size table such as Krejcie and Morgan’s (1970) for estimating proportions in a finite population, or the minimums recommended for your analysis method. Report the number you approached, the number who responded and the response rate. Our guide to how many participants a dissertation needs covers the options in detail.
Step 4: Choose or build your instrument
Whenever possible, use an existing validated scale rather than writing your own questions. Validated instruments come with evidence that they measure what they claim, and examiners trust them. For each instrument, report the author, year, number of items, dimensions and response scale, and evidence of reliability, ideally in South African samples. Our guide on where to find a validated questionnaire lists sources.
If you must write items yourself (for example, demographic questions or a context-specific section), pilot them with a small group from your population, check that respondents understand them as intended, and adjust them before the main collection. Translations into isiZulu, Sesotho, Afrikaans or other languages need a documented translation and back-translation process.
Step 5: Establish validity and reliability
- Content validity: do the items cover the construct? Expert review and alignment with the literature support it.
- Construct validity: do the items behave as the theory predicts? Factor analysis is the usual evidence in larger samples.
- Reliability: do the items consistently measure the same construct? Cronbach’s alpha is the most reported statistic; our guide to reporting Cronbach’s alpha shows how to present it.
Report reliability for your own sample, not only the values from the original development study.
Step 6: Plan the analysis before you collect data
Decide your statistical tests while writing the proposal. Matching tests to hypotheses in advance prevents the most common examiner criticism: analysis that does not answer the research questions.
| Question type | Typical test (parametric / non-parametric) |
|---|---|
| Describe levels of a variable | Means, standard deviations, frequencies |
| Compare two groups | Independent-samples t-test / Mann-Whitney U |
| Compare three or more groups | One-way ANOVA / Kruskal-Wallis |
| Relationship between two continuous variables | Pearson correlation / Spearman correlation |
| Predict an outcome from several variables | Multiple regression |
| Association between two categorical variables | Chi-square test |
Parametric tests make assumptions about your data, such as approximate normality. Check them and report what you found. If data are not normally distributed, see what to do when your data fail a normality test. For a full decision guide, see which statistical test to use in your dissertation.

Step 7: Ethics for quantitative studies
Surveys look low-risk, but ethics committees still expect detail: informed consent (often on the first page of an online survey), voluntary participation with a right to withdraw, anonymity or confidentiality, gatekeeper permission from the organisation, secure storage of data files, and how the data will be destroyed or archived. If you survey your own colleagues or students, explain how you reduced pressure to participate, for example by having someone else distribute the invitation.
Writing up quantitative results
A results chapter presents findings; the discussion chapter interprets them. Structure the results in the order of your research questions or hypotheses:
- Response rate and sample profile (demographics table).
- Reliability of each scale in your sample.
- Descriptive statistics for each main variable.
- Assumption checks for your tests.
- Results for each hypothesis: the test, the statistic, the df where relevant, the p-value and an effect size, plus a plain-language sentence.
An illustrative results sentence: “Work engagement was negatively correlated with turnover intention, r = −.42, p < .001, a moderate relationship; H1 was supported.” (Illustrative values.) Avoid pasting raw software output; rebuild tables in your faculty’s format. Our guide to turning SPSS output into a results chapter shows how.
Worked example: from research question to test (illustrative)
Seeing the whole chain for one study shows how the decisions connect. All details below are illustrative.
| Element | Decision |
|---|---|
| Research question | Is there a relationship between financial literacy and saving behaviour among final-year students at one Gauteng university? |
| Variables | Financial literacy score (independent); self-reported monthly saving (dependent); bursary status and field of study (controls) |
| Design | Cross-sectional correlational survey |
| Population and sample | Final-year undergraduates; stratified random sample by faculty from the registrar’s list, size justified by a power analysis |
| Instruments | A published financial literacy measure with reported reliability, plus demographic and saving items piloted with ten students |
| Analysis | Descriptive statistics; Spearman or Pearson correlation depending on distribution; multiple regression with controls |
| Ethics | Consent on the first survey page; anonymous responses; gatekeeper letter from the registrar |
Each row answers a question an examiner will ask. If you can complete this table for your own study before collecting data, your methodology chapter is largely planned.
When to add a qualitative strand (mixed methods)
Some questions need numbers and explanations. A survey can show that engagement predicts turnover intention; interviews with a few respondents can explain why. That combination is a mixed methods design, and it comes in recognised forms:
- Explanatory sequential: survey first, then interviews to explain surprising results.
- Exploratory sequential: interviews first to understand a topic, then a survey built from what you learned.
- Convergent: both strands at the same time, compared in the discussion.
Mixed methods is attractive but doubles the work: two instruments, two analyses, two sets of quality criteria and an integration section. For a coursework master’s or honours research report with a fixed deadline, a well-executed single-method study usually earns more than an over-stretched mixed design. Discuss it with your supervisor before committing.
Collecting the data: online, paper or both
Most South African survey dissertations now collect data online, but the choice affects who responds. Online surveys are cheap and quick and export straight to your analysis software, yet they under-represent people with limited data or connectivity, which matters in rural clinics, schools and community settings. Paper questionnaires reach those respondents but need capturing, checking and secure storage. Many studies use both, which is acceptable if you report it and check whether the two groups differ. Whatever you use, plan reminders (one or two polite follow-ups usually lift response rates), keep the questionnaire as short as your hypotheses allow, and test the full survey on a phone before launch, because many respondents will answer on one.
Common quantitative research mistakes examiners flag
- Research questions that cannot be answered with the planned test.
- Causal claims from a cross-sectional survey.
- Generalising from a convenience sample without acknowledging the limitation.
- Using an instrument without reporting its source, items and reliability.
- Reporting p-values without effect sizes.
- Running many tests and reporting only the significant ones.
- Tables copied straight from software with unexplained labels.
FAQ
What is quantitative research in simple terms?
It is research that measures variables with numbers and uses statistics to describe them, compare groups or test relationships, so that findings can be checked and, with the right sample, generalised.
What is the most common quantitative design for a master’s dissertation?
A cross-sectional survey using validated questionnaires, analysed with descriptive statistics and correlational or comparative tests.
How big should my sample be?
Justify it with a power analysis, a published sample size table or the minimum recommended for your analysis, and report your response rate.
Can I write my own questionnaire?
You can, but validated scales are preferred. If you write items, pilot them and report how you checked validity and reliability.
Do I need to report effect sizes?
Yes. A p-value tells you whether an effect is likely to be non-zero; an effect size tells you how large it is, which is what readers need.
Does Tesify run my statistics?
No. Tesify helps you structure your dissertation chapter by chapter. Your analysis and your writing stay yours, 100% written by you.
Structure your quantitative dissertation with Tesify
A quantitative dissertation is easiest to defend when questions, hypotheses, instruments and tests line up chapter by chapter. 50,000+ students have written 230,000+ chapters with Tesify, which helps you structure your dissertation chapter by chapter, and every word stays 100% written by you. Structure your dissertation with Tesify.
Sources cited
- Krejcie, R.V. & Morgan, D.W. 1970. Determining sample size for research activities. Educational and Psychological Measurement, 30(3):607-610. https://doi.org/10.1177/001316447003000308
