Handling Bias

January 2, 2015 • John Vandivier

Selecting against bias is a form of selection bias.
Jury selection may combat preconceived stereotypical notions, but this is effectively a bias against such notions, which are sometimes true.
Random selection is the only way to handle bias. Biases in opposite directions naturally cancel each other out. Bias \"washes out\" of large random samples. If there is any remaining bias, it is a bias in the direction of the common opinion of the market. This bias may be wrong, but we have reason to think that most subgroups of the population, including individuals, are more likely to be wrong than the whole market. There are a few exceptions, but there is no strong evidence even for these exceptions, which are the intelligent, the moral, the powerful, the rich, the naturally able, and the otherwise successful or better.
The deviations of the tendencies of any sample relative to a random sample may be taken as probabilistic indications of the characteristics of the nonrandom sample, characteristic, or direction. For example, if randomness of income is reduced then the deviations of the characteristics of the new sample may be taken as probabilistic indicators of the way in which randomness of income is reduced. If the way in which randomness of income is reduced is to exclude the poor, then emergent patterns may be taken as more likely to belong to the rich, or the non-poor to be more precise.