Hi! I wanted to better understand lines 57 and 58 from Day3’s practical regarding the inclusion of age as a covariate.
The lines are:
defAge ← mxMatrix( type=“Full”, nrow=1, ncol=1, free=FALSE, labels=c(“data.age”), name=“defL” )
pathB ← mxMatrix( type=“Full”, nrow=1, ncol=1, free=TRUE, values=0.01, label=“b11”, name=“beta” )
Thanks!
- Can you explain what each portion in the parenthesis is referring to. For instance, what is “data.age”?
OK. Consider defAge. type="Full" says that it is a “full” matrix, which can be of any dimensions, and in which every element may have a unique value. nrow=1, ncol=1 says that it will have 1 row and 1 column. free=FALSE because it does not contain any free parameters. In the MxModel’s namespace, it will be named “defL”. That leaves the tricky part, labels=c("data.age").
In an MxModel’s namespace, “data” always refers to the MxModel’s MxData object. Also, OpenMx interprets a period as meaning “member of”, a la C/C++. Thus, “data.age” refers to the variable damed “age” in the dataset.
Assuming the rows of the dataset are sampled independently, the -2logL of the full dataset is the sum of the -2logLs across all rows. That is, OpenMx evaluates the -2logL for each row, and sums them up to get the total -2logL. When OpenMx evaluates the -2logL for each row, the MxMatrix named “defL” will be populated with the value of the “age” variable specific to that row. Thus, the model-expected means for each row are conditioned upon each row’s value of age.
- Is pathB for the covariate?
Yes. It contains the regression coefficient from the regression of the phenotype onto age.