Statistical Bias: Survey Sampling

In survey sampling, bias is the tendency of a sample statistic to systematically over- or under-estimate a population parameter which cannot be accounted for by statistical methods (i.e., normality) since the data is inherently skewed

Selection Bias

Undercoverage: If this survey form was the only means of obtaining data, then people who do not have access to internet are inadequately represented

Nonresponse bias: If mail was distributed to account for people without access to internet, then respondents can differ in meaningful ways since it generally takes more effort to mail than it does to click submit

Voluntary response bias: If this survey was mailed to a subscriber list, the response from the list would not be indicative for critical feedback

Response bias

Leading questions: A question phrased for a desired answer

Social desirability: Responses will tend to skew towards social norms when questions are confidential

Can you think of other examples of bias? For further reading, refer to:

• Observer bias
• Survivorship bias
• Cause-effect bias (post hoc ergo propter hoc)
• Funding bias
• Cognitive bias