It depends.
The purpose of writing accessor methods (mutators etc.) is not just the blind
formal replacement of direct attribute access with mutators. The purpose of
writing accessors is to provide interface that decouples the clients of the
class from the physical implementation details of the corresponding attribute.
The benefits of this decoupling apply to other classes as much as they apply to
the class itself. For this reason the reasonable strategy to follow is to use
accessor methods whenever you can, and use direct attribute access only when you
must.
When you are writing methods that belong to the same class as the attribute in
question, the choice depends on the nature of the functionality under
consideration. If this is a very low-level functionality that is supposed to
have knowledge of/rely on the implementation details of the class, then you
might choose the access the attribute directly (for the sake of efficiency). If
this is a higher level functionality, you might be better off using the accessor
methods.
I always had this notion in my head like calling the method would be
cleaner than modifying the attributes directly. Mostly so that later
on you could add things that might be needed to the method without
changing anything else. So many SetText() always calls UpdateScreen()
you know?
The reason why I asked this time though was I am in a situation where
I would be adding multiple lines of checks to in the mutator to
account for the fact that I'm only initializing the attribute in this
case and wouldn't want subsequent functions calls like UpdateText() to
take place versus just setting the attribute directly in my Init().