What you'll learn
The method you choose determines what you can find out, and CAPE Caribbean Studies expects you to justify a choice rather than default to a questionnaire. This topic covers primary and secondary data, quantitative and qualitative approaches, the four main primary methods — questionnaires, interviews, observation and focus groups — with the strengths and weaknesses of each, the sampling techniques that determine whether findings can be generalised, and the concepts of validity, reliability, pilot testing and triangulation. It is the most directly examinable topic in Module 3 and the one your Internal Assessment methodology section is marked against.
Key terms and definitions
Primary data — data the researcher collects directly for the study at hand.
Secondary data — data collected by someone else for another purpose: census figures, published research, official reports, newspaper archives.
Quantitative data — data in numerical form, allowing counting, comparison and statistical treatment.
Qualitative data — data in the form of words, meanings and descriptions, allowing depth of understanding.
Population — the entire group the research concerns.
Sampling frame — the list from which a sample is actually drawn, such as a school register.
Sample — the subset studied.
Random sample — every member of the population has an equal chance of selection.
Stratified sample — the population is divided into groups and members selected from each, usually in proportion, so all groups are represented.
Systematic sample — every nth case is selected from a list after a random start.
Quota sample — the researcher fills set numbers from named categories, without random selection.
Purposive sample — cases are chosen deliberately because they have characteristics the study needs.
Snowball sample — existing participants recruit further participants; used where a population is hard to reach.
Convenience sample — whoever is available is used. Quick, and the weakest for generalisation.
Validity — whether an instrument measures what it claims to measure.
Reliability — whether it produces consistent results when repeated.
Response rate — the proportion of distributed instruments returned usable.
Pilot study — a small trial run of an instrument before full use.
Triangulation — using more than one method or source to check findings against one another.
Ethical consent — informed, voluntary agreement to take part.
Core concepts
Primary versus secondary data
Primary data is collected for your specific question, so it fits exactly, but it costs time and access. Secondary data already exists, is usually cheaper and can cover periods or populations you could never reach, but it was gathered for someone else's purpose and may not match your definitions, your period or your population.
Most good IAs use both: secondary sources establish context and support the literature review, while primary data answers the specific research question. Where you use secondary data, check when it was collected and how the categories were defined, because a mismatch between their definitions and yours is a genuine limitation that should be stated.
Quantitative and qualitative approaches
Quantitative methods produce numbers, permitting comparison, measurement of extent, and generalisation where the sample allows. They answer how many, how often and how much.
Qualitative methods produce words and meanings, permitting depth, nuance and understanding of reasons. They answer why and how it is experienced.
Neither is superior; they answer different questions. A study asking how many students hold a view needs quantitative data; a study asking why they hold it needs qualitative. Many strong studies use both, and saying so with justification is stronger than defending one approach against the other.
Questionnaires
A set of written questions distributed to respondents, usually the most practical method for a school IA.
Strengths. They reach many respondents quickly, produce standardised data that is straightforward to compare and tabulate, can be completed anonymously — which improves honesty on sensitive matters — and remove interviewer influence.
Weaknesses. Response rates are often low. Respondents may misread questions with no one present to clarify. Closed questions restrict answers to the options provided, so an unanticipated view is simply lost. They reveal little about reasons. And they exclude those who cannot read them comfortably.
Question design matters and is examinable. Closed questions produce quantitative data and are quick to analyse; open questions produce qualitative depth but are harder to compare. Avoid leading questions, which suggest an answer; double-barrelled questions, which ask two things at once so the answer cannot be interpreted; ambiguous wording; and jargon. Likert scales are useful for measuring attitudes on a consistent scale.
Interviews
A conversation in which the researcher asks questions directly.
Structured interviews follow a fixed schedule, giving comparability across respondents. Semi-structured interviews follow a guide but allow follow-up, which is usually the most productive form for a student project. Unstructured interviews are led by the respondent and produce the greatest depth and the least comparability.
Strengths. Depth and detail; ambiguity can be clarified immediately; follow-up questions can pursue unexpected answers; non-verbal cues are available; and they suit respondents who would not complete a written form.
Weaknesses. Time-consuming, so sample sizes are small. Interviewer effect: respondents may adjust answers to what they think the researcher wants. Analysis of transcripts is laborious. And anonymity is impossible, which matters on sensitive topics.
Observation
Watching behaviour directly rather than asking about it.
Participant observation means joining the group being studied; non-participant means watching without taking part. Either may be overt, with the group aware, or covert, without their knowledge — and covert observation raises ethical problems around consent that make it inappropriate for a school IA.
Strengths. It records what people actually do rather than what they report doing, which matters because the two often differ. It captures behaviour in its natural setting and can reveal things respondents would not think to mention.
Weaknesses. The observer effect: people behave differently when watched. It cannot access motivations or beliefs, only actions. Observations may be interpreted subjectively. It is time-consuming and usually covers few cases.
Focus groups
A guided discussion among a small group, typically six to ten people.
Strengths. Participants respond to one another, producing ideas an individual interview might not surface; efficient for gathering several perspectives at once; useful for exploring shared attitudes.
Weaknesses. Dominant participants can suppress others; people may not disclose sensitive or minority views in front of peers; the group may converge on a consensus that misrepresents individual positions; and moderating well requires skill.
Sampling
Sampling determines whether findings can be generalised, and it is where student projects most often overreach.
Probability methods — random, stratified, systematic — give every member of the population a known chance of selection, so findings can be generalised to the population with a stated confidence. They require a sampling frame.
Non-probability methods — quota, purposive, snowball, convenience — do not, so findings describe the sample and cannot be generalised. They remain legitimate where a sampling frame is unavailable or the population is hard to reach, provided the limitation is stated.
Stratified sampling is often the strongest realistic choice for a school study: dividing by form group or by sex and selecting proportionally from each ensures the sample reflects the population's structure rather than whoever volunteered.
Sample size matters, but representativeness matters more. A well-stratified sample of forty is more useful than a hundred respondents who happened to be in one corridor.
Validity, reliability and triangulation
Validity asks whether you measured what you claimed to. A questionnaire on attitudes to reading that actually measures willingness to appear studious is not valid.
Reliability asks whether repetition would give consistent results. An ambiguous question is unreliable, because respondents interpret it differently each time.
A pilot study on a small number of respondents before full distribution catches ambiguity, leading questions and instructions that do not work, and is one of the cheapest ways to improve an instrument.
Triangulation checks findings using more than one method or source. If questionnaire responses, an interview and school records point the same way, confidence rises; where they conflict, the conflict is itself a finding worth reporting.
Worked examples
Example 1: Justifying a method
Question: "A student wishes to investigate why some students in a school do not participate in extra-curricular activities. Recommend a suitable data collection method and justify your choice." (10 marks)
Outline. The question asks why, which calls for reasons rather than counts, so a qualitative method is indicated. Recommend semi-structured interviews: a guide ensures the key areas are covered while follow-up questions pursue reasons the researcher had not anticipated — which matters here, since the causes may be financial, domestic or social and could not all be predicted in advance. Justify further: ambiguity can be clarified on the spot, and respondents who would not complete a written form may still talk. Acknowledge the trade-off: small sample, no anonymity, laborious analysis. Then strengthen the design by adding a short questionnaire to establish how widespread each reason is once interviews have identified them, and note that this triangulates the two sources. A recommendation that names limitations and mitigates them scores far higher than one that does not.
Example 2: Evaluating an instrument
Question: "Identify two faults in the following questionnaire item and rewrite it: 'Don't you agree that the government's poor handling of the economy and rising crime has made life harder?'" (8 marks)
Outline. First fault: it is leading — "don't you agree" and "poor handling" tell the respondent the expected answer, which biases responses and undermines validity. Second fault: it is double-barrelled, asking about the economy and about crime in one item, so an answer cannot be interpreted; a respondent agreeing may mean either or both. A rewrite separates and neutralises: "How has the cost of living changed for your household over the past year?" with a rated scale, and a separate item on perceptions of crime. State that the rewrite removes the steer and isolates one variable per question.
Example 3: A short sampling question
Question: "Distinguish between a random sample and a stratified sample." (4 marks)
Outline. In a random sample every member of the population has an equal chance of selection, usually by drawing from a complete sampling frame; it is unbiased but may by chance under-represent a small group. In a stratified sample the population is first divided into groups — by form, sex or district — and members are selected from each, usually in proportion to their share of the population, which guarantees every group is represented. The distinction is that stratification deliberately controls the composition of the sample rather than leaving it to chance.
Common mistakes and how to avoid them
Defaulting to a questionnaire without justification. The method must follow the question. A study asking why is poorly served by closed questions.
Claiming generalisability from a convenience sample. If you surveyed whoever was available, your findings describe those respondents. Say so; it is a limitation, not a failure.
Listing strengths and weaknesses without applying them. Questions usually ask you to recommend and justify for a specific study, so the evaluation must be attached to that case.
Ignoring the observer effect and interviewer effect. Both are standard evaluation points and both are frequently omitted.
Confusing validity with reliability. Validity is measuring the right thing; reliability is measuring consistently. An instrument can be reliable and invalid — consistently measuring the wrong thing.
Proposing covert observation. It cannot secure informed consent and is inappropriate for a school study. Say why if you are asked to evaluate it.
Omitting the pilot study. It is cheap, improves the instrument and is creditable in a methodology section.
How this links to your Internal Assessment
Your methodology section is marked directly against this material: the method chosen, why it suits the question, the sampling technique and its justification, sample size, and how the instrument was developed and tested.
Three things reliably earn marks. Pilot your instrument and say what the pilot changed. State your sampling technique by name and explain why it was appropriate given the frame available to you. And in your limitations, be specific rather than formulaic — "a convenience sample of thirty students from two Form 5 classes means findings cannot be generalised beyond this school" is worth far more than "the sample was small".
Consent and confidentiality belong here too. Obtain permission from the school, explain the purpose to respondents, and make participation voluntary.
Exam technique for data collection methods
Match method to question type. How many and how often call for quantitative methods; why and how experienced call for qualitative ones.
Always justify, never merely name. A recommendation earns marks through its reasons and its acknowledged trade-offs.
Learn the sampling techniques precisely, with one line each on how they work and whether they permit generalisation. They are among the most frequently examined items in Module 3.
Use the question-design faults — leading, double-barrelled, ambiguous, jargon — as a checklist. Questions asking you to critique an item almost always contain one or two of them.
Propose triangulation where the study allows it. Suggesting a second method to check the first demonstrates design thinking rather than method recall.
Quick revision summary
Primary data is collected for your study and fits exactly but costs time and access; secondary data already exists but was gathered for another purpose and may not match your definitions or period. Quantitative methods answer how many and permit generalisation; qualitative methods answer why and permit depth. Questionnaires reach many respondents and give standardised, anonymous, comparable data but suffer low response rates, restricted closed answers and no insight into reasons. Interviews — structured, semi-structured or unstructured — give depth and allow clarification but are time-consuming, non-anonymous and subject to interviewer effect. Observation records what people actually do rather than what they report but cannot access motives and is subject to the observer effect; covert observation cannot secure consent and is inappropriate for a school study. Focus groups surface ideas through interaction but risk domination and self-censorship. Probability sampling — random, stratified, systematic — permits generalisation and needs a sampling frame; non-probability sampling — quota, purposive, snowball, convenience — does not, and the limitation must be stated. Validity is measuring the right thing, reliability is measuring consistently, a pilot study catches faults cheaply, and triangulation checks findings across methods.