Intermediate level · Module 1: Data Collection and Survey Data · Lesson 2 of 3
Sampling and Survey Weights
Understand common sampling designs and why weights are needed to make survey estimates represent the population.
In this lesson you will
- Describe simple random, stratified and cluster sampling
- Apply survey weights to calculate estimates
Key concepts
stratified samplingcluster samplingsampling weightweighted estimate
The lesson
Simple random sampling gives everyone the same chance. Stratified sampling selects separately within groups such as districts, guaranteeing enough cases in each. Cluster sampling selects villages first and then households, which is cheaper in the field but less precise.
When some groups are deliberately over- or under-sampled, each response carries a weight: the number of population households it represents. Unweighted averages then give the wrong population picture.
Calculate weighted estimates with np.average(values, weights=w) or survey-aware libraries, and report how the sample was designed.
Step by step
- Read the survey's sampling description.
- Compare unweighted and weighted percentages.
- Explain the difference.
- Write the methods note.
Practice activity
Calculate weighted indicators for the provided survey and explain why they differ from unweighted ones.
Expected result: Correct weighted estimates and a clear explanation.
Check your understanding
-
Why might a survey over-sample a small district?
Show answer
To have enough cases to estimate that district reliably.
-
What does a survey weight represent?
Show answer
How many population units a response stands for.
Taught live by an instructor — online or in person. Free course lessons are for self-study and do not lead to a certificate — see registration options.