ecological fallacy

The error of substituting a correlation or property computed on aggregated groups for the corresponding correlation or property at the individual level, on the assumption the two must agree. W. S. Robinson's 1950 paper is the mathematical demonstration the fallacy is named for — the term itself is reported (via secondary sources, not independently confirmed by this vault) to have been coined slightly later, by Selvin (1958).

Matters to this vault as the aggregation-specific instance of the general "cheap scalar substituted for a harder property" pattern: it directly names the statistical mechanism behind Garfield/Seglen's warning that a journal's mean citation rate cannot substitute for an individual article's citation count (see claim-robinson-1950-ecological-fallacy-names-garfield-seglen-mechanism and claim-garfield-seglen-and-steck-warnings-share-a-scalar-proxy-structure), predating Garfield's own warnings by roughly two decades. Sits alongside construct validity as the older ancestor of a pattern this vault keeps finding rediscovered field by field.

References

• claim-robinson-1950-ecological-fallacy-names-garfield-seglen-mechanism