Interpret >1 and negative results from variance parametrisation of twin model

Hi All,

I am using the new parametrization for a bivariate model, as suggested in "Type I Error Rates and Parameter Bias in Multivariate Behavioral Genetic Models".

The results (see below) include an rA correlation that is higher than one.

Am I doing something wrong or I have to interpret the rA correlation as 1?

The phenotypic correlation is .2. Is it be correct to say that is due to 1, -.96 and .2 for A, D and E respectively?

There are also negative values: there is no way to interpret these results, right?

Thank you so much in advance

$V
mxAlgebra 'V' 
$formula:  VA + VD + VE 
$result:
           [,1]       [,2]
[1,] 0.26015570 0.08651586
[2,] 0.08651586 0.64434885
dimnames: NULL
 
$iSD
mxAlgebra 'iSD' 
$formula:  solve(sqrt(I * V)) 
$result:
          [,1]      [,2]
[1,] 1.9605744 0.0000000
[2,] 0.0000000 1.2457746
dimnames: NULL
 
$rA
mxAlgebra 'rA' 
$formula:  solve(sqrt(I * VA)) %&% VA 
$result:
          [,1]      [,2]
[1,] 1.0000000 3.1877784
[2,] 3.1877784 1.0000000
dimnames: NULL
 
$rD
mxAlgebra 'rD' 
$formula:  solve(sqrt(I * VD)) %&% VD 
$result:
            [,1]        [,2]
[1,]  1.00000000 -0.95732896
[2,] -0.95732896  1.00000000
dimnames: NULL
 
$rE
mxAlgebra 'rE' 
$formula:  solve(sqrt(I * VE)) %&% VE 
$result:
           [,1]       [,2]
[1,] 1.00000000 0.20271077
[2,] 0.20271077 1.00000000
dimnames: NULL
 
$US
mxAlgebra 'US' 
$formula:  cbind(VA, VD, VE, VA/V, VD/V, VE/V) 
$result:
            VA          VA           VD           VD          VE          VE         SA          SA         SD          SD
US 0.036015032 0.130102084  0.053615331 -0.096971671 0.170525336 0.053385447 0.13843645 1.503794611  0.2060894 -1.12085427
US 0.130102084 0.046249649 -0.096971671  0.191371989 0.053385447 0.406727210 1.50379461 0.071777344 -1.1208543  0.29700059
           SE         SE
US 0.65547415 0.61705966
US 0.61705966 0.63122206

Hope that you are all well during this difficult situation