rates were only of a medium level, so the sample
may not have been representative of parents at the
school. Second, the sample was not likely to have
been representative of families of children with
autism generally because these families were relatively
highly educated. Third, the sample did not
contain parents of children with other developmental
disabilities. Finally, many of the children
boarded at the school during the week and were
only home on weekends and during school holiday
periods. These points, and those raised earlier,
should be considered as we explore the implications
of the findings. Most important is that the
results need to be replicated with larger and more
representative samples.
The issue of the residential status of a number
of the children (at least as weekly boarders at the
school) is one that could have been predicted to
be important in the context of the present study.
There were no differences on Hospital Anxiety
and Depression Scale scores between mothers and
fathers whose children lived with them and those
whose children boarded at the school. However,
it was possible that parents whose children resided
at the school would be less affected by their
child's behavior problems than were other parents.
Although not our main focus, and thus not
reported in the Results section, we repeated the
hierarchical regression analyses but explored residential
status as a moderator variable in place of
self-efficacy. There was no evidence that residential
status moderated the impact of child behavior
problems on parents in the manner suggested.
Thus, it seems unlikely that the results relating to
self-efficacy can be explained by the child's residential
status.
Two further methodological issues are worthy
of note. The first is that the regression models for
mothers were able to explain more of the variance
in anxiety (52%) and depression (62%) than were
the regression models for fathers' anxiety (39%,
excluding fathers' age) and depression (27%, excluding
fathers' age). This serves again to support
the argument that children with disabilities and
associated parental psychological factors, such as
self-efficacy, are likely to affect fathers and mothers
differently.
The final methodological issue relates to the
problem of source variance and potential measurement
overlap. Although we avoided the
source variance problem of parents reporting on
their child and on themselves, respondents did
report on both their self-efficacy and their mental