The random-effects assumption says nothing about why effects differ from one another effects may differ from one another in either fixed-effects or random effects models. The real substance of the random-effects assumption is that the differences among effects, whatever their source, are 'natural', not having been controlled or contrived by the research design, and are independent of other effects. That is, the effects in the data represent a random sample of the effects in the population. Independence implies that knowing the value of a particular cp;k, for example, is of no help in predicting the values of other business-unit effects or the values of any industry, corporate, or year effects. An important exception to this assumption, involving an association between industry and corporate effects, is discussed below.