Finally, we acknowledge some obvious limitations of our study. One involves the restriction required for our operationalisation of family type.
Jenkins and Zunguze(1998) argue that the difference between single parent and
two parent families is inadequate. Additional classification criteria might include the custody arrangement following divorce as described above or the presence of step- or
halfsiblings.
Secondly, as our dataset is cross-sectional, we cannot make definitive assessments about the direction of the effects.