Research, data interpretation and development skills
This domain tests the practical skills behind the other eight: designing a small study, choosing a sample, reading tables and graphs, and judging whether a claim is supported by evidence. Questions typically present data and ask what can and cannot be concluded.
Research design
- A workable sequence: research question, hypothesis, method, data collection, analysis, conclusion. A hypothesis must be testable.
- Primary data you collect: questionnaires, interviews, observation, transects, counts. Secondary data comes from censuses, World Bank and UN sources, or NGO reports - always check date, definition and who collected it.
- Quantitative data gives numbers and patterns; qualitative data explains reasons and perceptions. Using both is triangulation and strengthens conclusions.
- Ethics: informed consent, anonymity, no harm, honest reporting.
Sampling
- Random sampling gives every member an equal chance, using random numbers.
- Systematic sampling takes every nth name on a list; stratified sampling divides the population into groups and samples proportionally.
- Quota and opportunity samples are non-random: quick but biased. A larger sample reduces sampling error, but representativeness matters more than size.
Working with numbers
- Mean is distorted by outliers; median is the middle value; mode is the most frequent.
- Percentage change = (new − old) ÷ old × 100. A move from 20% to 25% is +5 percentage points but a 25% relative increase.
- Index numbers set a base year at 100 so change over time is easy to compare.
Worked example
A school of 900 students needs a 10% stratified sample by year group. Year 1 has 270 students, so sample 270 × 0.10 = 27 from that group; repeat for each group and check the total equals 90. If instead you interview the first 50 students through the gate at 8am, that is an opportunity sample - cheap, but it over-represents early arrivers and misses absentees and latecomers, so findings cannot be generalised.
Traps that cost marks
Truncated graph axes exaggerating small changes; claiming causation from correlation when a third factor explains both; leading questionnaire items such as "Don't you agree the clinic is badly run?"; quoting percentages from a sample of 15; ignoring non-response and missing data.