Reliability, validity and representativeness ask different questions about research. Reliability concerns consistency in the relevant measurement or procedure. Validity concerns whether the evidence appropriately captures what the study intends to investigate. Representativeness concerns how the sample reflects the population relevant to the claim. A study can have a strength in one area and a limitation in another.
To use these concepts well, identify the feature of the actual design and explain its effect. Avoid writing that a method is simply “good” or “bad” without showing why. This guide uses fictional school-club examples for O Level Sociology practice. It does not present invented findings as real research or assume that a method's name alone determines the quality of every study using it.
Begin with three separate questions
For reliability, ask whether the relevant procedure produces consistent measurements under comparable conditions. For validity, ask whether the question, indicator or interpretation fits what is being investigated. For representativeness, ask who is included and how that group relates to the wider population. Keeping those questions separate helps you avoid attaching the wrong concept to a limitation.
For example, interviewing only club members about the reasons all students do or do not join raises an inclusion issue. That is different from asking an ambiguous attendance question. The first concerns the sample's relationship to the population; the second can affect what the answer means. Both matter, but explaining them as the same problem weakens the analysis.
Explain reliability through the procedure
Suppose two observers count participation in a club meeting. One counts every brief comment, while the other counts only contributions longer than a minute. Their totals may differ because they use different operational definitions. Agreeing on clear recording categories and training can improve consistency in the observation procedure.
Do not assume that every difference between repeated results proves unreliability. Participants' behaviour or circumstances may genuinely change. The relevant question concerns comparable conditions and the process used to measure. A study repeated in another year may encounter a changed social setting, so its differences need interpretation rather than an automatic label of failure.
Recognise consistency without assuming accuracy
A procedure can be consistent while measuring the wrong thing. Imagine asking every student the identical question, “Is the club good?” and coding yes as proof of high participation. The wording may be applied consistently, but approval is not the same as attendance. Reliability of administration would not resolve the mismatch between indicator and research aim.
This distinction explains why standardisation is useful but insufficient. Identical questions and coding can support comparison, yet ambiguous language or an unsuitable indicator can remain. When evaluating, state what the design improves and what it does not establish. Avoid saying that a reliable procedure automatically produces valid conclusions about every question.
Explain validity through the intended concept
Suppose the research aim is to understand why students feel included in a club. A simple count of meetings attended may provide relevant context but may not capture their sense of inclusion. A student can attend frequently and still describe feeling excluded. The indicator and the intended concept are related but not identical.
A design might need questions or other evidence that explores the experience more directly. Explain that requirement in the context. Do not define validity only as “the results are true” and leave it there. Ask whether the evidence and interpretation support the particular claim, and whether important meanings or alternative explanations have been overlooked.
Inspect question wording
Leading, vague or double-barrelled questions can affect how well responses represent the intended views. “Why is our excellent club so useful?” pushes an assumption into the question. “Is the club enjoyable and useful?” combines two judgements that may differ. “Do you attend often?” leaves the frequency category unclear.
Improve the design by defining the period, separating distinct issues and using suitable wording. A pilot can reveal misunderstanding, although it cannot guarantee that every later answer is accurate. The research-methods guide explains how design choices affect the evidence. Keep the evaluation attached to a specific question feature rather than calling every questionnaire invalid.
Consider the research interaction
Participants may respond differently depending on who asks and where the conversation occurs. A student discussing club difficulties in front of the club leader may be reluctant to speak openly. That possibility concerns the interaction and the meaning of the responses, not merely the number of people interviewed.
Use the scenario to explain the likely issue and an appropriate design consideration. Do not assume every interview produces dishonest answers. Rapport, privacy, wording and participant understanding can affect the process. A useful answer distinguishes a potential limitation from a proven description of what every participant actually did.
Define the population before judging representation
Representativeness cannot be assessed without knowing which population the study concerns. A sample of current club members may be suitable for a question about those members' experiences. It is more limited for a claim about all students' reasons for non-participation because non-members are excluded.
State the target group precisely. “Students” can mean a year group, school, region or another defined population. A finding about one selected group does not automatically describe every student elsewhere. The sociology concepts guide provides related advice on using specific contexts and avoiding unsupported universal claims.
Examine who can enter the sample
Look at the sampling frame, selection process and access. If an online survey reaches only people using one club messaging group, students outside that group may never have the opportunity to respond. A large response count does not automatically repair that exclusion. Ask how the included people may differ from those missing.
Non-response can also matter. Even with a suitable initial selection, the final responding group may differ if some participants do not answer. Explain the relevant possibility without assuming its direction unless evidence is supplied. A good evaluation identifies the design feature and its implication for the claim rather than relying only on the word “biased”.
Use stratification examples carefully
Suppose a fictional population of 200 students contains groups of 100, 60 and 40 under a relevant category. A proportional sample of 20 could allocate 10, 6 and 4 places. This matches those proportions, but the category must be relevant to the study and selection within groups still needs an appropriate procedure.
Matching one characteristic does not prove that the sample represents every important feature. Non-response and incomplete lists can remain. Do not turn a correct proportional calculation into a guarantee of perfect representation. Explain what the design attempts to preserve and which uncertainty remains.
Compare method strengths without automatic rankings
A structured questionnaire may support consistent administration and numerical comparison. A flexible interview may explore meanings in greater depth. These are potential strengths of particular designs, not proof that one method always wins on every quality criterion. The research aim determines which evidence is needed.
For a study of unfamiliar experiences, follow-up discussion may reveal meanings that fixed options miss. For a comparison of clearly defined responses across a group, standardised items may be useful. Discuss the trade-off in context. Avoid writing that qualitative evidence is automatically valid or quantitative evidence automatically reliable without examining the actual collection and interpretation process.
Explain how combined evidence can help
A study might combine attendance records with interviews. Records can describe the recorded pattern, while interviews explore accounts of participation. The sources can complement each other when their definitions and periods are clear. Their combination does not automatically remove every limitation in either source.
If the sources disagree, investigate the reason. A register may use a different definition of attendance, or an interview may refer to another period. Do not choose the source that best fits a preferred conclusion without examination. Explain how the additional evidence addresses a specific uncertainty and what remains unresolved.
Write an evaluation as a chain
Name the feature, explain its effect and connect it to the relevant quality concept. For example: only current members were invited, so non-members' experiences are excluded, limiting the sample's ability to represent all students for this research aim. That chain is clearer than “the sample is bad and invalid”.
The command-word guide and mark-scheme review guide explain how to match development to the task. Use a relevant limitation rather than listing every concept mechanically. The answer should show why this feature matters for this claim.
Practise applying all three separately
Take one fictional design and write a short paragraph about each concept. Identify a different feature where appropriate. Then change the research aim and examine whether the judgement changes. A sample suitable for describing members may be limited for explaining non-members, even when the same people and method are used.
Students seeking guidance can explore the O Level Sociology course, course catalogue and admissions FAQ. Bring a specific design and your evaluation. Feedback can then identify whether the difficulty is the concept's meaning, the selected evidence or the connection between them.
Frequently asked questions
Can research be reliable but have limited validity?
Yes. A procedure can be applied consistently while using an unsuitable indicator or question. Explain what is consistent and why the evidence may not capture the intended concept. Reliability alone does not settle validity.
Does representativeness mean a large sample?
Size is only part of the issue. The population, sampling frame, selection and non-response also matter. A large group drawn from one narrow source may not support claims about a broader population.
Are qualitative methods always more valid?
Do not assume that from the label alone. A design can explore meanings in depth, but wording, interaction and interpretation still need evaluation. Match the evidence to the particular research aim and explain the relevant strengths and limits.
Does a different repeated result prove low reliability?
Not automatically. Behaviour, circumstances or the population may genuinely change. Consider whether conditions and procedures are comparable before attributing the difference to inconsistency in measurement.
How can I avoid mixing the three concepts in an answer?
Ask three separate questions: is the procedure consistent, does the evidence capture the intended issue, and how does the sample relate to the population? Then connect a specific design feature to the appropriate concept through a clear explanation.
Ask about O and A Level courses
Share your subjects, current study stage and the support you need. The Abeel Eschool team can help you check the current course options.