[R] Computing tetrachoric covariance matrices for multiple imputed datasets using MICE package

[Ian McPhail]

Using the mice package, I have created multiple imputed datasets to deal
with missing data. I am looking for an example of the R code to use in
order to analyze the set of imputed datasets using tetrachoric correlations
in such a way that after pooling, I will have a combined tetrachoric
covariance-variance matrix to use as input for an exploratory factor
analysis. I have taken a few attempts at the with() command in the mice
package, using the poly() function, but do not quite know what I'm doing so
am out of my depth with the R code. All of the examples for using the
with() command in the mice package involve lm() and regression formula.

I provide some examples below of what I have attempted, but I think my
question is about not understanding what the expression part of the with()
function code is about and how to implement different analysis onto the
imputed datasets using the with() function.

For example, using an example with only 3 variables, I have attempted the
following code,

imp<- mice(df, meth = pmm, m = 25)

fit <- with(imp, poly(var1, var2, var3))

Alternatively, I have tried:

imp<- mice(df, meth = pmm, m = 25)

fit <- with(imp, poly(var1:var3))

Alternatively, I have tried:

imp<- mice(df, meth = pmm, m = 25)

fit <- with(imp, poly(imp))

I have attempted the same series of code, but using the polychoric()
function in the psych package.

The data I am working with are 22 scale items that have a yes/no response
type. I am not very savvy with R, but I appreciate any help people are able
to provide.

tia,

Ian

	[[alternative HTML version deleted]]