Best-shot aggregation makes large values dominate small ones by design

Evidence

One of the generic concerns with OLS as a statistical technique is the influence of “outliers.” The idea is that these outliers have “too much” influence over the final parameter estimate. This is because, as one is moving in the best shot direction, larger values substitute more and more for smaller ones and have a bigger and bigger impact over the final value. While this is seen as negative in an OLS context, the best shot logic suggests that it might be quite reasonable, depending on the substantive content of the concept. — Gary Goertz, Social Science Concepts and Measurement, 1014

Categorization

• SEED: Angle: the outlier problem in statistics is secretly an aggregation-logic choice — what looks like a nuisance is a substantive decision about how dimensions substitute.
• XREF: connects to: A simple powerful-group count proxies well for full resource-weighted interest-group indices — Martin Gilens and Benjamin I. Page, Testing Theories of American Politics:…, 59