Primary/secondary concerns whether data are newly generated for the present study or existing material reused. Quantitative/qualitative concerns numerical measurements/counts or material such as accounts, images and field notes used to understand meanings.
These dimensions cross: a new online numerical survey is primary and quantitative, while archived interview accounts reused are secondary and qualitative. A method or internet location alone does not determine every category.
Cross origin and data form
Choose a method for the research question
Questionnaires may combine fixed options and own-word answers; structured interviews use a schedule and can be remote or in person, while semi/unstructured interviews allow differing degrees of follow-up. No method automatically guarantees detailed, truthful or comparable answers.
Observation can be participant/non-participant and overt/covert on separate dimensions, with counts or qualitative notes. Experiments examine effects through a designed comparison; causal inference depends on control and design, not merely the label experiment.
Match the method to the evidence needed
Distinguish population, frame and achieved sample
The target population is whom the question concerns; the sampling frame is the operational source for selection; the selected sample and the people who actually respond can differ. Missing entries and non-response can exclude relevant groups.
Simple random selection gives eligible frame units an equal chance under that design, but does not guarantee a representative response. Systematic selection uses an interval after an appropriate start; list order or periodic patterns can matter.
Trace who the findings concern
Compare stratified, quota and referral recruitment
Stratified probability sampling divides the population into strata and samples within each. Proportional allocation is one option; quota recruitment also fills subgroup numbers but normally selects non-randomly within them.
Snowball recruitment uses participant referrals and can help reach contacts, while excluding disconnected networks. Opportunity/convenience samples use accessible people; neither a large sample nor matching one characteristic removes every selection bias.
Compare selection procedures
Apply original fictional proportional allocation
Original library-access population: 120 Year 9 and 180 Year 10 pupils, total 300. For a proportional sample of 30, allocate 120/300 × 30 = 12 to Year 9 and 180/300 × 30 = 18 to Year 10, then select randomly within each defined stratum.
If 12 of the selected 30 reply, the response rate is 40%, not a representative result merely because selection started randomly. Investigate whether missing pupils differ in relevant access needs; these practice figures are not a real school survey.
Calculate original fictional sample allocation
Pilot neutral and accessible questions
A pilot is a small-scale trial of recruitment, wording, timing and analysis before the main study. Original poor item: How much do you love our brilliant library? A more neutral item asks How often did you use the library in the last school week? with clear options.
Add an optional own-word question about barriers to explore meanings, but avoid unnecessary personal information. Ambiguity, leading language, double questions and unavailable response options can damage the intended measure.
Improve the original library question
Separate reliability, validity and generalisation
Reliability concerns consistency under comparable conditions; validity how well the intended concept/inference is captured. A repeatedly identical but misleading question can be reliable without measuring the intended concept well.
Representativeness concerns relevant characteristics of a stated population. Generalisation must consider coverage, response, measurement and uncertainty; an association does not prove causation, and triangulation can reveal disagreement as well as strengthen interpretation.
Separate research-quality questions
Plan consent, privacy and withdrawal practically
Participation should be voluntary and appropriately informed, with accessible information, power differences considered and clear withdrawal arrangements. Children’s research needs suitable school/adult support and context-specific ethical procedures rather than assumed consent from attendance.
Confidentiality protects entrusted information under explained limits; anonymity concerns identifiability and is not the same promise. Safeguarding or legal limits and practical limits after anonymisation should be clear; do not secretly collect sensitive information for a classroom task.
Make ethical safeguards operational
Evaluate a design through specific trade-offs
For the fictional library question, fixed responses can show patterns and interviews may explore why access is difficult. Existing opening-time records can add context, but their definitions and relevance need checking; mixed methods do not automatically guarantee truth.
Weigh practical access, time/cost, consistency, validity, sampling and ethical safeguards against the aim. State what the achieved evidence supports and what remains unknown, revise the plan if needed and avoid claiming to have conducted fictional research.
Evaluate the fictional library design
スライド
Sign up free to view the lesson slides
Step through every slide for this topic — plus flashcards and revision notes — with a free account.
