4.2.2.1.2 Using Multivariate Logistic Regression Coefficients to Make Estimation
4.2.2.2 Allocation towards Daughter .1 Analysis of Logistic Regression
For testing the allocation towards the daughter, the sample size consist total of 465 respondents.
In the chi-square statistic and its significance level, the probability of obtaining this chi-square statistic (9.637) if there is in fact no effect of the independent variables, taken together, on the dependent variable. In this case, the model is statistically not significant because the p-value is more than 0.05 significant levels.
Table 4.9 Variables in the Equation for Daughter
Variables B S.E Exp(B) Result(Sig.)
Age -0.010 0.014 0.990 No
Health Level -0.042 0.068 0.959 No
Gender -0.217 0.202 0.805 No
Education (Primary) -0.514 0.307 0.598 Significant at 10 per cent
level.
Education (Secondary and above)
-0.422 0.308 0.656 No
Ethnic 0.149 0.205 1.161 No
Marital 0.205 0.244 1.227 No
Constant 1.241 1.067 3.459 No
Source: Developed for Research
In table 4.9, B represents the values or coefficient for the logistic regression equation for predicting the dependent variable from the independent variable.
They are in log-odds units. The logistic equation is
Log (p/1-p) = 1.241 – 0.010 Age – 0.42 Health level + 0.149 Ethnic + 0. 205 Marital Status – 0.514 Education (Primary) – 0.422 Education (Secondary and above) – 0.217 Gender
From the above result, education level is significant at 10 per cent level whereas the remaining independent variables do not have significant results.
Although overall model in the chi-square result shows the model is not significant, the Table 4.9 shows that Education (primary) is statistically significant at 10 per cent level. This means that there is a significant difference in the distribution of wealth towards the daughter among those who obtain until primary school and non-schooling. Since there are three categories of education, there are two set of coefficient, odds ratio, and significant value for education. Each set represents the comparison between one education category and the reference category. Those who are non- schooling are the reference category. The odds ratio for the Education (Primary) is the comparison between those who study until Primary with non- schooling. The result indicates that 0.598 of people who has only primary school of education is less likely to distribute their wealth towards their daughter as non-schooling. Likewise, who has the education level of secondary school and above has 0.656 less likely to distribute their wealth towards their daughter in comparing with those who are non-schooling.
Health level is statistically not significant. From the above result, the odds that the elderly married non-Muslim will more likely to leave the wealth towards their daughter is 0.959 times higher for each one unit increase on the health level as shown in the Exp(B) column in Table 4.9.
Gender is also shows not significant. This means there is no differences between male and female in allocating their wealth towards their daughter.
Odds ratio for gender is 0.805.
Age is not significant and finds no evidence that age is related to the distribution of wealth towards daughter. The negative relationship (-0.010) revealed in the coefficient column (B) shows that the older the people are, the less likely to allocate their wealth towards their daughter. The odd ratio column (Exp(B)) indicate that for each additional year of age of the people, his or her odds of distributing the wealth towards their daughter is only 0.990.
Ethnic (whether is Chinese or Indians) is not significant and found no evidence has any differences in distributing the wealth towards daughter.
Marital status (whether is currently married or divorce or widowed) is not significant and found no evidence has any differences in distributing the wealth towards daughter.
4.2.2.2.2 Using Multivariate Logistic Regression Coefficients to Make Estimation.
Since the education level (primary school) is statistically significant, therefore the variables will be used to make prediction towards the wealth distribution pattern in daughter.
Estimation 1: Estimation of the likelihood that a person who obtains non- schooling (coded as 0) is more likely to distribute their wealth towards daughter. The regression is form by using the mean for age group (60 years old), health level in average (4), and the most common value for others independent variables.
The coefficient has drawn for this equation from the logistic equation output.
Log-odds = 1.241 – 0.010 Age – 0.42 Health level + 0.149 Ethnic + 0.
205 Marital Status – 0.514 Education (Primary) – 0.422 Education (Secondary and above) – 0.217 Gender
Odds = Exp (1.241 – 0.010 Age – 0.42 Health level + 0.149 Ethnic + 0. 205 Marital Status – 0.514 Education (Primary) – 0.422 Education (Secondary and above) – 0.217 Gender)
Probability = odds/ 1+odds Probability =
Probability =
Probability = 26.10%
A person who has non-schooling has 26.10% probability of the likelihood in distributing the wealth towards daughter.
Table 4.10 Sources of the numbers in the above equations Constant = 1.241
Independent Variables B Value
Age -0.010 60 (Mean)
Health Level -0.420 4 ( Average Health Level)
Gender -0.217 0 (Female)
Education (primary) -0.514 0 (Non-schooling)
Education (secondary and above) -0.422 0 (Non-schooling)
Ethnic 0.149 0 (Chinese)
Marital Status 0.205 0 (Currently Married)
Source: Developed for Research
Estimation 2: Estimation of the likelihood that a person who Primary school (coded as 1) is more likely to distribute their wealth towards daughter. The regression is form by using the mean for age group (60 years old), health level in average (4), and the most common value for others independent variables.
The coefficient has drawn for this equation from the logistic equation output.
Log-odds = 1.241 – 0.010 Age – 0.42 Health level + 0.149 Ethnic + 0.
205 Marital Status – 0.514 Education (Primary) – 0.422 Education (Secondary and above) – 0.217 Gender
Odds = Exp (1.241 – 0.010 Age – 0.42 Health level + 0.149 Ethnic + 0. 205 Marital Status – 0.514 Education (Primary) – 0.422 Education (Secondary and above) – 0.217 Gender)
Probability = odds/ 1+odds Probability =
Probability =
Probability = 17.50%
People who obtain until primary school level of education have 17.50%
probability of the likelihood in distributing the wealth towards daughter.
Table 4.11 Sources of the numbers in the above equations Constant = 1.241
Independent Variables B Value
Age -0.010 60 (Mean)
Health Level -0.420 4 ( Average Health Level)
Gender -0.217 0 (Female)
Education (primary) -0.514 1 (Primary)
Education (secondary and above) -0.422 0 (Non-schooling)
Ethnic 0.149 0 (Chinese)
Marital Status 0.205 0 (Currently Married)
Source: Developed for Research
From the Estimation 1 and prediction 2, the probabilities shows not much differences (26.1% and 17.5%) as the overall equation is not statistically significant in distributing the wealth for daughter.