I am using regular Mx (unfortunatly i cannot find an active forum), and would like to clarify the meanings of the output received after running in an ACE with an ordinal moderator script.

1. Are the estimates of T,U,V the moderated components? (often denoted by betas)

2. If the estimate for a certain parameter is not zero, yet the CI contains zero- does this imply that this component is not significant?

3. To obtain the percentages of trait variance for additive genetics, should I simply square the estimate (also, does the answer to this question depend on the significance/non significance if the moderated component of A)?

This is an example for the output of Mx:

6 Confidence intervals requested in group 1

Matrix Element Int. Estimate Lower Upper Lfail Ufail

A 1 1 1 95.0 0.2197 0.1865 0.2368 0 0 0 0

C 1 1 1 95.0 0.0000 -0.1044 0.1044 0 0 0 0

E 1 1 1 95.0 0.1475 0.1339 0.1636 0 0 0 0

T 1 1 1 95.0 0.0063 -0.0230 0.0288 0 0 0 0

U 1 1 1 95.0 0.0000 -0.0825 0.0825 0 0 0 0

V 1 1 1 95.0 -0.0071 -0.0209 0.0082 0 0 0 0

Hello Lia

2 - yes. Be aware that the symmetric confidence intervals around zero are likely erroneous on the negative side - the algorithm has found a lower than zero lower CI that gives the same likelihood as the one that is positive. However, this would be because the model expectations only contain the square of the estimate.

3 - No, probably not. If a path has, say (a + bx) where x is the moderator, the expected variance is a^2 + b^2x^2 + 2abx

At least, when the model is that the same factors are involved in the change with the moderator.

Thank you for your reply!

I attached the script.

1, yes, T U and V are the moderated components.

The model is perhaps NOT identified though because the variables are binary. See Medland, S.E., Neale, M.C., Eaves, L.J., Neale, B.M. (2009). A note on the parameterization of purcell's G x E model for ordinal and binary Data. Behavior Genetics, 39 (2): 220-229.

Note that whenever you see the same -2lnL fit, but

differentparameter estimates, this is strong evidence that the model is not identified. You can also use mxCheckIdentification to test models.