mice impute missing values? Output of complete always has NAs. about mice HOT 2 CLOSED

You may skip removal of collinear variables at initialization by the remove_collinear flag.

mic <- mice(res$amp, diagnostics = FALSE , remove_collinear = FALSE)

If you do so, mice tries to resolve the problem during the iterations. See the large list of complaints in

mic$loggedEvents

We are working on numerically stabler methods that can handle these difficulties.

from mice.

Your data are almost collinear

> cor(res$amp, use = "pair")
          x         y
x 1.0000000 0.9998434
y 0.9998434 1.0000000

mice tries to be smart (and evade some nasty computational problems along the way), and removes nearly collinear variable beforehand. You can see that this happens at mic$loggedEvents.

mic$loggedEvents
  it im co dep      meth out
1  0  0  0     collinear   y

So this is expected behaviour, but I can also see why you didn't expect it :-)
Stef.

from mice.