Selection Bias

Notes

Claim

Selection Bias is when we unintentionally create a situation of Output quality depends on input quality by having biased data to use in our model or in our thinking process.

Explanation

Why it matters

Examples

For example, deciding to run a "do you support gun legislation" survey at a NRA conference will lead to biased results and won't truly represent the actual population Population and sample estimates.

Some cases can be less obvious, like doing surveys over the phone, while not asking ourselves which population is less likely to have a phone (Amish, elderly, kids...).

Supporters

Opposers

Open questions

Visual