Hypothesis 1, which concerned the degree of
interdependence within elder-caregiver dyads on
spirituality and psychological well-being, was
tested using the intraclass correlation coefficient
(ICC). Figure 1 indicates that arrows a and b refer
to the relationship of spirituality and psychological
well-being between elders and caregivers. The
ICC was used to assess the interdependence of
each trait within elder-caregiver dyads. The ICC
provides a measure of the amount of the total
variance in a single trait that is attributable
to group membership. In this case, a larger ICC
indicates a greater degree of interdependence in
scores between elders and their caregivers for a
particular variable.
Hypotheses 2 and 3 guided the examination of
actor effects and partner effects. The model of
actor and partner effects of spirituality on psychological
well-being was examined using the
psychological well-being index as the dependent
variable and the spirituality index as the independent
variable. Analyses were conducted using
APIM linear mixed modeling, which allows for
the relationship between variables to be examined
even when the data structure is highly interdependent
within clusters (see Campbell & Kashy,
2002 for an explanation of the application of
linear mixed modeling to dyadic data). Dyads
were treated as interdependent clusters with a
random intercept.Aseries of analytical steps were
followed to test the study hypotheses.
First, the variance in psychological well-being
remaining after controlling for the model variables,
measured by the full model Bayesian Information
Criterion (BIC), was compared with the
total uncontrolled variance in psychological wellbeing,
measured by the null model BIC (i.e., the
model including only the random intercept, omitting
all fixed effects, and representing the total
between-dyad variability in the data). A significant
change in BIC indicated that the overall
model accounted for more than a chance proportion
of the difference between individuals’
well-being. Second, the main effect for actor
spirituality (hypothesis 2) was examined. A significant
actor effect indicates that an individual’s
spirituality is positively associated with their own
psychological well-being. Third, the main effect
for partner spirituality (hypothesis 3) was
examined. A significant partner effect indicates
that an individual’s well-being was positively
associated with their partner’s spirituality. Finally,
two interaction effects were examined: actor
effect by dyad role, and partner effect by dyad
role. A significant actor effect by dyad role interaction
indicates that there is a significant