06.5.5 · Skill in Statistical reasoning
Sampling and selection bias
When the data comes from a group that does not represent everyone.
Survey respondents, power users and early adopters are not typical users. Survivorship bias, such as only studying users who stayed, is common in product data. Asking who is missing from the data prevents many wrong conclusions.
Resources
1- articleStart hereHow survivorship bias affects your analysis ↗Data School · 2019 · 4 min
A short explanation with business examples of drawing conclusions only from the users who stayed.
Tools
none yetTools for this topic are being researched.