The 1 reference context in paper Richard Sperling, Christopher F. Baum () “Multivariate portmanteau (Q) test for white noise” / RePEc:tsj:stbull:y:2001:v:10:i:60:sts19

  1. Start
    If we can make a guess at these two factors, that is, intraclass correlation and cluster size, we can adjust our sample size calculations by multiplying the variance of the sample size formulas by the following correction
    (Kish 1965)
    +-(m-1) wherem=average number of observations per cluster,-=intraclass correlation If there is no intraclass correlation, that is,-=0, this correction factor reduces to 1, thus causing no increase in the sample size estimate.
    (check this in PDF content)