At this juncture, given the posts presented thus far, I wish to re-visit and argue for the statement: most. Some Ranters take issue with "most" because it doesn't recognize the individual.

However, for the sake of expediency, "most" is more efficient than dealing with the individual. We simply can not argue on a case-by-case situation the merits of the individual.

Take the fields of psychology, sociology, anthropology and mathematics. These fields use "trends" (nearly all, most, etc) to predict outcomes.

Various models developed within these disciplines are very accurate and spot on.

Simply because one individual out of 100 doesn't fit the model, doesn't make the model incorrect or wrong - ergo generalization can be applied.

Contrarian, extraordinaire