ECON1012 · Data Analytics
Data & Sampling
Data & Sampling (Module 1, Week 1) sets up the language the whole of ECON 1012 runs on. You learn what statistics is — a way to get information from data — and the split between descriptive and inferential statistics that structures the course. The core vocabulary is population, sample, parameter and statistic, plus census versus sample and the 3 Vs of Big Data. You classify data as numerical, ordinal or nominal — which determines what calculations are legal — and distinguish cross-sectional, time-series and longitudinal (panel) data. Sampling plans (simple random, stratified, cluster) and survey methods bring in response rates and the crucial contrast between sampling error, which falls as sample size grows, and non-sampling error, which does not. Finally, you summarise nominal and ordinal data with frequency tables, bar charts and pie charts.
What this chapter covers
- 01Data → Statistics → Information: descriptive vs inferential statistics
- 02Population · sample · parameter · statistic — the four-word vocabulary every question starts with
- 03Census vs sample; Big Data's 3 Vs — Volume, Velocity, Variety
- 04Data types: numerical (quantitative) → ordinal (ranked) → nominal (categorical), and which calculations each allows
- 05Sampling plans: simple random · stratified · cluster (including multistage cluster)
- 06Sampling error (falls as n rises) vs non-sampling error (does not)
- 07Cross-sectional vs time-series vs longitudinal (panel) data
- 08Charts for nominal/ordinal data: frequency table → relative frequency → bar chart · pie chart (y% → y × 3.6°)
Population, sample, sampling plan and a pie-chart slice
- 2 marks(a) Population = all loyalty-program members of the chain; sample = the 1,200 members surveyed at the 6 selected stores.
- 2 marks(a) Parameter = the true proportion of ALL members who would use the service — the claimed 40% refers to this. Statistic = the sample proportion p̂ = 456/1200 = 0.38, i.e. 38%.
- 2 marks(b) Cluster sampling: a simple random sample of whole groups (stores) is taken, then every member inside the chosen clusters is surveyed. It is NOT stratified — stratified would draw a random sample of members from every one of the 60 stores.
- 2 marks(c) Relative frequency = count ÷ total: would use 456/1200 = 0.38; unsure 264/1200 = 0.22; would not 480/1200 = 0.40. Check: 0.38 + 0.22 + 0.40 = 1.00 ✓
- 1 mark(d) 100% ↔ 360°, so y% ↔ y × 3.6 degrees: the 'would use' slice = 38 × 3.6 = 136.8°.
Key terms
- Population
- The group of all items (data) of interest; a census measures every one of them, which is accurate but costly and often impractical.
- Sample
- A set of items drawn from the full population — the affordable route to information about the population, at the price of sampling error.
- Parameter
- A descriptive measure of a population (a true mean μ or proportion p); fixed but usually unknown — the thing inference tries to pin down.
- Statistic
- A descriptive measure of a sample (x̄, s, p̂); it varies from sample to sample and is used to draw conclusions about the parameter.
- Cluster sampling
- A simple random sample of whole groups (clusters), surveying every element inside the chosen groups; useful when the population is dispersed or no full list exists, but it may increase sampling error.
- Sampling error
- The gap between a sample result and the population value caused purely by which observations happened to be selected; it shrinks as n grows — unlike non-sampling error (acquisition errors, non-response, selection bias), which does not.
Data & Sampling FAQ
How do I tell a parameter from a statistic in ECON 1012 questions?
Ask what group the number describes. A parameter describes the population — it usually hides behind a claim about 'all' customers, products or voters, and is normally unknown. A statistic is whatever you compute from the sample (x̄, s, p̂). In the classic exam scenario, the manufacturer's claimed percentage is the parameter and the tested batch's percentage is the statistic.
What is the difference between stratified and cluster sampling?
Stratified sampling divides the population into mutually exclusive strata and draws a simple random sample from EVERY stratum. Cluster sampling takes a simple random sample of whole groups and surveys everyone inside the selected groups only. Watch for multistage descriptions (pick zones, then blocks, then survey all households) — they read like stratified but are multistage cluster sampling.
Does a bigger sample fix every kind of error?
No. A larger n reduces sampling error only. Non-sampling error — data-acquisition mistakes, non-response error and selection bias — comes from how the data are collected, so the fix is better survey design and follow-up, not more observations. This contrast is a favourite quiz question.
Do I need a textbook or Excel for this topic?
There is no prescribed textbook for ECON 1012 — the official outline states no learning resources are required, and the myLearning modules are the material. Excel is introduced in workshops for handling data, but the final exam is hand-calculation, so practise the Week 1 classifications and chart computations on paper. Check the current unit outline and myLearning for anything offering-specific.
Studying with AI? Sia — free AI economics tutor works through ECON 1012 step by step.
Exam move
Treat Week 1 as vocabulary training, not maths. Drill the four-word chain — population, sample, parameter, statistic — until you can label any scenario in seconds: a claim about 'all customers' is a parameter; anything computed from the survey is a statistic. The classic trap is stratified versus cluster: stratified samples within every group, cluster randomly picks whole groups — and multistage descriptions are often worded to look stratified when they are cluster. Only a larger sample reduces sampling error; non-sampling error (non-response, selection bias) needs better design, not more data. Know the data-type hierarchy (numerical → ordinal → nominal) and the two chart recipes: relative frequency = count ÷ total, pie angle = % × 3.6. Then rehearse with the re-attemptable Module 1 quiz on myLearning.
