|
Message/Author |
|
|
This is probably very straightforward, but why don't R2 and error equate to 100%? If I input: TITLE: preliminary play; DATA: FILE IS martin.dat; VARIABLE: NAMES ARE Beh INt Att SN3 SN2 SID Mnm SEf PBC AffN AnR Fam Aff2; USEVARIABLES ARE Beh INt att mnm affn sn3 SID anr sef; CATEGORICAL IS beh; MODEL: beh on int; int ON att sn3 sid sef anr; att on affn mnm sef sid; output: MODINDICES(0) standardized; savedata: DIFFTEST IS Francebase.dat; The residual variances I get are: INT 0.364 ATT 0.445 and the R-Squares are: R-SQUARE Observed Residual Variable Variance R-Square BEH 0.742 0.703 INT 0.861 ATT 0.512 So with Beh (for example) how does the residual variance related to R-Square? Any help much appreciated. |
|
|
They sum to 1 only when you consider the residual variances in the standardized version. |
|
|
Thank you for that - it makes sense now. Regards Barbara |
|
Back to top |
|
|