[R-sig-ME] very basic HLM question

Paul Johnson pauljohn32 at gmail.com
Tue Feb 8 19:36:23 CET 2011

2011/2/7 Douglas Bates <bates at stat.wisc.edu>:
> 2011/2/7 Sebastián Daza <sebastian.daza at gmail.com>:
>> I have run a one-way anova random effect in SPSS and I get that the
>> between-component variance is 59.004. With this figure I can get the same
>> ICC that I obtained using ICC1. I haven't found a way to get that
>> between-component variance using R.
> I don't know how to reproduce the SPSS results in R.

This strikes me as the wrong question.  Find out what formula SPSS
uses.  So far, in this thread, you've claimed that you found one
formula for ICC in an article, and you referred to the ICC1 function
from multilevel which uses a different formula.  We still don't know
exactly how SPSS is making its calculation.

The formulae that R routines use are in plain sight for you to
inspect.  If SPSS would offer you the same courtesy, then we would not
be left in this weird situation of trying to guess what they mean by
some number.

Paul E. Johnson
Professor, Political Science
1541 Lilac Lane, Room 504
University of Kansas

More information about the R-sig-mixed-models mailing list