Page 55 - Contributed Paper Session (CPS) - Volume 4

CPS2119 Noor Azlin M. S. et al.
likelihood (-1268.5) as compared to probit link-function (-1294.5), clog-log
link-function (-1373.6) and nlog-log link function (-1301.5). This is also
supported by the smallest AIC value for ordinal regression model with logit
link function (2593.0) as compared to ordinal regression model with other link
functions; probit link function (2645.0), clog-log link function (2658.9) and
nlog-log link function (2803.2). Therefore, this finding suggested that the
ordinal regression model with logit link function is the best fit model as
compared to model with probit, clog-log and nlog-log to represent the level
of family well-being data.
These results show that all social factors are identified as significant factors
to determine the level of family well-being except safety. In contrast,
comparing other models with different link function, the demographic factors
such as ethnic is found to have a significant relationship with the level of family
well-being. Interestingly, the significant parameter estimates are all positive,
which means, each of the category of the contributed factors are more likely
to have a higher-level satisfaction of family well-being.
TABLE 6. Performances ordinal regression model with different built-in link functions
Ordinal Model with Akaike Information
Selected Link Functions Log-Likelihood Criteria
Logit -1268.5 2593.0
Probit -1294.5 2645.0
Clog-log -1373.6 2658.9
Nlog-log -1301.5 2803.2

5. Conclusion
The ordinal regression model with logit link function is found to be the
best model for describing the relationship between the level of family well-
being and the covariates which include ethnic, family relationship, economic
situation, health status, community relationship, religion practice and housing
and environment. It is clear that social determinants are the significant factors
for explaining the level of family well-being. However, it is a bit surprising to
see that demographic factors except for ethnic and socioeconomic factors are
not significant. This could possibly be due to the Malays which represent the
major race in the country who are usually easily get contented with the little
things that they have.

Acknowledgement
This article uses sample data from the Malaysian Family Well-Being Index
Survey (MFWIS) 2011 conducted by the National Population and Family
Development Board Malaysia (NPFDBM).

44 | I S I W S C 2 0 1 9