Okay now I can reproduce the issue. Weird. I can only guess that it's a side-effect of how Acrobat/Reader handily tries to fit the data to the form. When it could not find the node that you asked for, it looked up the tree to find another match.
Here's another approach that seems to work.
var oSecLevelXML = XMLData.parse(xfa.record.toplevel.seclevel.saveXML(), false);
var oIDNode = XMLData.applyXPath(oSecLevelXML,"//seclevel/author/title/ID");
if (oIDNode == null)
xfa.host.messageBox("null");
else
xfa.host.messageBox(oIDNode.value);
Jared Langdon
www.jlangdon.ca