Page 49 - Contributed Paper Session (CPS) - Volume 4

CPS2119 Noor Azlin M. S. et al.

A comparison of ordinal regression in an analysis
of factors associated with family well-being
2
1
Noor Azlin Muhammad Sapri , Kamarulzaman Ibrahim
1 National Population and Family Planning Development Board of Malaysia (NPFPDBM), Kuala
Lumpur, Malaysia
1,2 School of Mathematical Sciences, Faculty of Science and Technology, Universiti Kebangsaan
Malaysia

Abstract
The aim of this study is to determine the factors which are significantly related
to the level of family well-being by using ordinal regression model. The data
for this study is obtained from the Family Well-Being Index Survey for the year
2011, conducted by the National Population and Family Development Board
of Malaysia. It involved a stratified random sampling of 2,808 households and
contains information on level of family well-being, demographic,
socioeconomic and social characteristics of each household. The ordinal
regression model involving four different link functions which include logit,
probit, clog-log and nlog-log are fitted to the data for determining factors
which have a significant relationship with the level of family well-being. Based
on Akaike Information Criteria (AIC), it is found that ordinal regression model
with logit link function provides the best fit for the data. The factors such as
ethnic, family relationship, economy situation, health status, safety, religion
practice, community relationship and housing and environment are
significantly related to family well-being.

Keywords
Ordinal Regression Model, Ordinal Data, Family Well-Being, Malaysia.

1. Introduction
Ordinal regression models are playing a central role in studies where the
researcher needs to quantify subjective variables. In the case of social sciences
research, ordinal regression models are now becoming a very powerful tool in
analysing data from questionnaires (Javali & Pandit 2010; Mouriño 2013). The
main idea is to model an ordinal response variable as a function of some
explanatory variables. In its basic formulation, the ordinal regression model
can be viewed as an extension of the logistic regression model for binary
response variables.
In Malaysia, little is known about the well-being of families since there is
not much research done in this topic especially at the national level. Malaysian
Quality of Life Index and Malaysian Youth Index are the two examples of local
study. But both of it did not stress on the family well-being. However, in 2011,

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