Jeremy,

Yes, the calculation of the ICC could be more a lot more straightforward in the SPSS Mixed Model package (and for that matter in the HLM software package). I run the unconditional model for the variable of interest (i.e, the one that simply has the lowest-level variable with group membership information). I divide the between group variance (Tau for the intercept) by the sum of the between and within group variance.

Cheers,

KS

---- Jeremy Miles <jnvm1@york.ac.uk> wrote: > I don't think I've ever seen a straightforward way to calculate the ICC > using SPSS. (In Stata it's very easy - using the loneway command.) Is > there a way to do it that I'm missing? > > Jeremy > > Statisticsdoc wrote: > > Keith Starborn > > www.statisticsdoc.com > > > > > > Dear Ryan, > > > > As Jeremy Miles suggests, your research question appears to be one that > > would be addressed by Hierarchical Linear Modeling (HLM), i.e., Mixed Models > > in SPSS. One important aspect of your research question appears to require > > an estimate of the degree to which data from the same family members is > > associated. To address this question, you can utilize the Intra-Class > > Correlation between subjects who are nested in the same unit (i.e., family > > members, students in the same classroom). HLM analyses provide the > > information that is needed for the Intraclass Correlation, and allow you to > > carry out appropriate tests of the degree to which variation in family-level > > independent variables is associated wirg differences in subject-level > > dependent variables. > > > > HTH, > > > > KS > > > > For personalized and professional consultation in statistics and research > > design, visit > > www.statisticsdoc.com > > > > E-Mail: statisticsdoc@cox.net > > > > > > > > -----Original Message----- > > From: SPSSX(r) Discussion [mailto:SPSSX-L@LISTSERV.UGA.EDU]On Behalf Of > > Ryan Black > > Sent: Saturday, June 10, 2006 11:19 PM > > To: SPSSX-L@LISTSERV.UGA.EDU > > Subject: Hierarchical/Nested Multiple Regression Question > > > > > > Hi, I have recently learned the idea of nested variables and I am > > having trouble figuring out how to incorporate a nested variable into a > > multiple regression analysis on SPSS. I understand the concept of > > having a full model versus reduced model (aka partial f or multiple > > partial f test) and looking at the Rsquare change (the variance of the > > DV attributed to the added predictor/predictors above and beyond the > > other variables in the model), but I do not see how that applies to > > having for instance relatives nested into their families. I would > > assume I would have to take into account the interdependence of family > > members. > > In terms of entering the data into the SPSS database, I believe I would > > add family as a variable (i.e. there are five members per family so I would > > dummy code this categorical variable). Is that correct? In terms of the > > multiple regression analysis, I > > would (as usual) enter the DV into the DV box, and the predictors in > > the IV box, but how would I take into account the effect of families > > since relatives are nested into their families and are therefore not > > independent? I do not believe the answer is to simply include family with > > the other predictors as the full model and compare that to the reduced > > model > > without family. Could somebody please explain to me how one would run a > > nested > > multiple regression on SPSS? Does it require the use of syntax? Any help > > would be greatly appreciated! > > Thank you, Ryan > > > > > -- > Jeremy Miles > mailto:jnvm1@york.ac.uk http://www-users.york.ac.uk/~jnvm1/ > Dept of Health Sciences (Area 4), University of York, York, YO10 5DD > Phone: 01904 321375 Mobile: 07941 228018 Fax 01904 321320 > > NOTE: New address from September 2006: > RAND Corporation, 1776 Main St, Santa Monica, CA, USA. > (New email and stuff too, but I don't know it yet).

-- For personalized and experienced consulting in statistics and research design, visit www.statisticsdoc.com