Lesson 4 of 8 · All lessons in this module
- Populations, samples and sampling error
- Stratified and cluster sampling
- Convenience and judgmental sampling, and choosing a method
- Sample size and mixing populations
- The central limit theorem
- The standard error of the sample meanVideo lesson
- Confidence intervals for the mean
- Bootstrap and jackknife resampling
Lesson 4 of 8 · 12 min
Sample size and mixing populations
More data reduces sampling error, but with diminishing returns, and only if every observation comes from the same population.
In short
- Larger random samples give smaller sampling error, but each extra observation helps less than the one before.
- A minimum sample size is needed for a given accuracy; beyond some size, extra data adds little.
- Stratified sampling tends to beat simple random sampling at small sample sizes; the advantage narrows as the sample grows.
- All observations must come from the same population (one distribution). Pooling data from different regimes or strategies gives a sample that represents no population.
- A smaller, homogeneous sample can be better than a larger, mixed one.
Unlock this lesson free for 7 days
Create a free account to get 7 days free: every lesson, video and flashcard, and the question bank. No card needed.