Sampling and Coverage Bias, explained.
Sampling bias arises when a selection or coverage process systematically distorts an estimate relative to its intended population.
Why it happens
Eighty online households include sixty supporters; twenty offline households include five. Online-only support is seventy-five percent, while total support is sixty-five percent; collecting more online responses does not cover offline households.
Sampling bias arises when a selection or coverage process systematically distorts an estimate relative to its intended population.
Read the result
Frame support is an exact fictional quantity here. A larger count within the same incomplete frame cannot add households the frame excludes.
A worked example
Count the households an online survey cannot reach
Choose the online-only frame and read support seventy-five percent among eighty households.
Reveal the offline group of twenty households with five supporters.
Combine sixty-five supporters across all one hundred households to distinguish population support from online-frame support.
OPTIONAL DEEPER DETAILGo deeper: inside the model
Inside this model
Eighty online households include sixty supporters; twenty offline households include five. Online-only support is seventy-five percent, while total support is sixty-five percent; collecting more online responses does not cover offline households.
Where this idea is useful
A practical use
Ask who cannot enter a survey before interpreting a large response count as representative.
A common misconception
“A very large response count guarantees a representative survey.”
Selection and coverage can remain systematically incomplete as the number of responses grows; more of the same frame does not remove the gap.
What this explanation leaves out
- All counts are invented and fully observed in the teaching ledger. Nonresponse, random-sampling variance and real population prevalence are not estimated.
Does this town estimate actual online survey bias?
No. The numbers are authored to isolate a coverage mechanism, without claiming prevalence or a measured effect in a real population.
Who has no route into the sample that supports your estimate?
Associated thinkers
Associations marked provisional are awaiting source review.
Further reading
Explore the original research or the teaching reference behind this experiment.