Node/data point graphics simply don't show points where any of the key fields are missing, but graphics made from elements completely fail if any field is not defined everywhere over the domain, whether it is the whole mesh or a subgroup of it.
It makes more sense for the domain of graphics to be the domain of the coordinate field (by default), then further restricted to the domain type and subgroup field.
Node/data point graphics simply don't show points where any of the key fields are missing, but graphics made from elements completely fail if any field is not defined everywhere over the domain, whether it is the whole mesh or a subgroup of it.
It makes more sense for the domain of graphics to be the domain of the coordinate field (by default), then further restricted to the domain type and subgroup field.