The first question was whether the Family disease Management Scale has
reliability or not. Reliability is obtained by examining Cronbach’s
alpha, composite reliability, and factor load coefficients.
Cutoffs for Cronbach’s alpha include more than or equal to .80 for a
suitable scale, .70 for an acceptable scale, and .60 for a scale for
exploratory purposes (Garson, 2016). As presented in Table 2, Cronbach’s
alpha is 0.883 for support, 0.894 for normalization, and 0.931 for
participation.
Cronbach’s alpha may over or undervalue scale reliability.
Thus, composite reliability may be preferred, which ranges from 0 to 1.
In a confirmatory model, composite reliabilities should reach .70 or
more than for an adequate model (Henseler, Ringle, & Sarstedt, 2012) .
Accordingly, the present study meets the composite reliability criterion
including 0.906 for support, 0.915 for normalization, and 0.942 for
participation (Table 2).
Cross-loadings mean each indicator variable has a higher correlation
with its latent variable than another latent variable (Garson, 2016). As
shown in Figure 1, structural path coefficients (loadings), which are
the path weights connecting the factors, was presented. The present
study could fulfill the criterion for discriminant validity.
The second question was whether the Family disease Management Scale is
valid or not. To this aim, the AVE index was examined.
The AVE presents the correlation of a construct with its indicators
(Henseler, Ringle, & Sarstedt, 2015). The average variance was 0.502
for support, 0.525 for normalization, and 0.62 for participation, which
meets adequate model criteria (Table 2).
Table 2. Reliability and
convergent validity