Individual and ecological correlation in a general context

This is a methodological study, applicable to all levels of living systems, which contrasts system wide measures with measures of individual units or components. If one samples K units and, in Unit i, samples ni observations of two characteristics, the correlation between these characteristics within a unit, the individual correlation, may be a different parameter from the correlation between summary measures of each unit, the ecological or systemwide correlation. If such a distinction is not taken into consideration in designing studies to evaluate relationships of characteristics of variable systems and in analyzing data from such studies, confusing and inaccurate inferences may be drawn. This general methodological point is illustrated by a study of testosterone level and orgasmic frequency in the human male.