Simple Random Sampling

How: Every item has equal chance of selection

When to use: Homogeneous population, no important subgroups

Example: Pick 1,000 customers randomly from 100,000 total

Pro: Unbiased, easy to understand
Con: May miss small but important groups

Stratified Sampling

How: Divide population into groups (strata), then sample from each

When to use: Population has distinct subgroups you care about

Example: Sample 200 users from each region (North, South, East, West)

Pro: Ensures representation of all groups
Con: Requires knowing group sizes upfront

Cluster Sampling

How: Randomly select groups (clusters), then sample everyone in those clusters

When to use: Population is geographically spread or naturally grouped

Example: Pick 10 random stores, survey all customers in those stores

Pro: Cost-effective, practical for large areas
Con: Higher variance than simple random

Systematic Sampling

How: Select every nth item (e.g., every 10th customer)

When to use: Processing ordered lists efficiently

Warning: Watch for periodic patterns in your data!

Common Sampling Biases

• Selection bias: Only sampling people who opt-in
• Survival bias: Only analyzing successful cases
• Non-response bias: People who don't respond differ from those who do