4. Findings
4.3 Hypothesis Testing
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
A university education is more important for a boy than for a girl (0.816), along with men make better business executives than women do (0.849) accounted for the highest factor loading.
Optimism (n = 2982)
The dependent variable included two components: feeling of happiness (M = 1.83, SD = 0.60) and state of health (M = 1.82, SD = 0.74). The extraction method
applied was principal component analysis. The KMO test reported a 0.66 an acceptable indicator of sampling adequacy. One component was extracted, explaining 68% and of the total variance. Feeling of happiness accounted for the highest factor loading (0.826).
Section 6.2 presented the descriptive statistic of the sample. Section 6.3 shows the statistical analysis per each hypothesis formulated in this research.
4.3 Hypothesis Testing
This research is aimed at finding associations between the independent variable educational level and the dependent variable Millennials’ attitudes. Furthermore, it analyses the interaction effect that an individual’s geographic origin (moderator variable) may have on the relationship of the dependent and independent variable. This section presents the results of each hypothesis tested and its respective statistical analysis.
Multiple regression and the logistic regression model for binary responses have been used to determine the aforementioned associations and interaction effects. The study of attitudes in Social Science is complex and contested, since individuals’
attitudes almost always have more than one reason. Hence, in order to provide alternative explanations for the association, four models were derived from the statistical analysis:
No interaction model 1: incudes independent variable and moderator variable —geographic origin.
No interaction model 2: model 1 plus the remainder of the controlled variables —background index.
‧
Complete model 4: model 3 plus the remainder of the interaction terms.
Hypothesis 1:
The higher the educational level, the less family dependence a Millennial will tend to have. Yet, East Asians Millennials are more likely to be family dependent then Westerns Millennials are.
Table 4-2: Logistic Regression Model on Family Dependence (n = 2,515)
Model 1 Model 2 Model 3 Model 4 Highest educational attained 0.040 0.022 1.041 -0.068 0.027 0.934
* -0.035 0.034 0.966 -0.047 0.040 0.954
Geographic Origin 1.044 0.086 2.840
*** 1.549 0.109 4.706
Social class (subjective) -0.174 0.060 0.841
** -0.179 0.060 0.836
Origin x Employment Status -0.079 0.112 0.924
Origin x Marital Status 1.133 0.297 3.104
Likelihood-Ratio 3270.889 2597.661 2594.933 2488.587
Cox and Cornel R2 0.69 0.288 2.88 0.318
Dependent variable: family dependency.
*p < .05. **p < .01. ***p<0.001
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
The significant negative association between educational attainment (IV) and family dependence (DV) appears in model 2 with no interaction. Hence, other variables controlled, it is likely that higher levels of education are associated with lesser probabilities of family dependence. The association loses significance in models 3 and 4 with interaction; that is, educational attainment does not affect directly family dependence.
Geographic origin (MV) has a significant positive association with family dependence along the four models. That is, East Asians are associated with higher probabilities —around 5 times greater— of family dependence than their Western counterparts.
Little evidence exits of interaction between origin (MV) and education (IV).
Evidence exist of interaction between origin and age, and it has a positive moderator effect on age. That is, East Asians as they age, they increase the likelihood that family dependence occurs.
To summarize, hypothesis 1 has been partially proved by the direct effect that origin (MV) has on family dependence. One can say, that education does not directly affect family dependence, but is an indirect cause through origin.
Hypothesis 2:
The higher the educational level a Millennial has, the more self-centered will tend to be. Yet, East Asian Millennials are more likely to take care of their elders than Western Millennials are.
This hypothesis measured attitudes considering the following dimensions:
respect to elders, elders considered a burden in society, older people getting too much from the government, and young people having more productivity at work than older ones have.
Exist a significant negative association between educational attainment (IV) and position toward family and elders (DV). The association holds along from the no interaction models to the interaction ones. Therefore, as a cohort, the higher the level of education a person has, the more self-centered he or she will tend to be.
‧
Geographic origin (MV) has a significant negative association with position toward family and elders (DV). That is, under statistical control, it is plausible that East Asians have a more self-centered attitude toward family and elders than their Western counterparts.
Origin x Employment Status -0.080 0.046
Origin x Marital Status 0.149 0.129
Origin x Children -0.135 0.067 *
Origin x Religion -0.055 0.087
Origin x Social Class -0.109 0.044 *
R2 0.080 0.094 0.097 0.104
F for change in R2 92.791 4.840 7.197 2.379
Dependent Variable: Position Toward Family and Elders p < .05. **p < .01. ***p<0.001
Nonetheless, exist evidence of an interaction between origin*education, which in turn has a positive moderator effect on educational attainment. That is, it is likely that in East Asian societies the higher educated a person is, the less self-centered he or she will tend to be.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Summarizing, the hypothesis has been disproved. On one hand, the interaction association between educational attainment, attitudes toward family and elders, and geographic origin is significant. Yet, the direction of the association is negative, contrary to what has been posed in hypothesis 2.
Hypothesis 3:
The higher the educational level a Millennial has, the more progressivist attitude toward non-traditional family roles will tend to have. Yet, East Asians Millennials are more likely to abide traditional family roles than Western Millennials are.
Hypothesis 3 has been measured with two components. The first one — traditional family roles— consist of a person’s attitudes toward abortion, homosexuality, sex before marriage and divorce. The second one consist of a person’s attitudes toward women’s roles in society in terms of their participation in politics, education, and enrollment in the labor force.
First component: traditional family roles
Exist a significant positive association between educational attainment (IV) and traditional family roles (DV) in the models with no interaction. That is, the higher level of education a person has, the more progressivist stance he or she will tend to have.
Geographic origin (MV) has a significant negative association with traditional family roles. That is, it is very likely that East Asians have a more conservative attitude toward traditional family roles, than their Western counterparts.
Exist evidence of strong interaction between origin*education. The negative association between origin and traditional family roles is somewhat tempered by the moderation effect of origin*education on origin. That is, among East Asians as they get more educated, the less conservative he or she will tend to be.
Summarizing, the first component of hypothesis three: attitudes toward traditional family role has been proved. Not only has educational achievement a direct effect on attitudes, but also have geographic origin and education an interaction effect on the relationship.
‧
Table 4-4: Regression Model on Traditional Family Roles (n = 2,326)
Second component: attitudes toward women’s role in society
Exist a positive significant association between educational attainment (IV) and attitudes toward women’s role in society (DV). The association holds significant along the four models. That is, the higher the educational level a person has, the less conservative attitudes his or her will tend to have.
The association between geographic origin (MV) and attitudes toward women’s role in society (DV) is significant and negative. Therefore, it is plausible that East Asians have a more conservative attitude than their Western counterparts.
Little evidence exists of the interaction origin*education. Hence, one can say that a person’s attitudes are better explained by the unique direct effect of education and geographic origin.
Origin x Employment Status -0.047 0.041
Origin x Marital Status -0.272 0.113
Origin x Children 0.047 0.058
Origin x Religion 0.351 0.078 ***
Origin x Social Class 0.017 0.040
R2 0.151 0.194 0.203 0.216
F for change in R2 206.880 17.616 25.713 5.636
Dependent variable: Traditional Family Roles
*p < .05. **p < .01. ***p < 0.001
‧
Table 4-5: Regression Model on Women’s Role in Society (n = 2,292)
Model 1 Model 2 Model 3 Model 4
Dependent Variable: Women’s Role in Society
*p < .05. **p < .01. ***p < 0.001
Hypothesis 4:
The higher educational level a Millennial has, the less optimistic will tend to be.
Yet, East Asians Millennials are more likely to have a greater sense of optimism than Western Millennials are.
The variable optimism was measured in terms of a person’s self-perceived happiness and self-perceived health.
Model one without interaction shows that the association between educational attainment (IV) and optimism (DV) is significant and negative. That is, the higher educated a person is, the less optimistic they will tend to be. However, the association loses significance after the statistic control in models with interaction.
‧
Exist a significant positive association between geographic origin (MV) and optimism (DV). The association holds along the four models. That is, it is very likely that East Asians have a greater sense of happiness and health, than their Western counterparts.
Little evidence exists of an interaction between origin*education. Hence, one can say that education does not directly affect optimism but is and indirect cause through geographic origin.
Table 4-6: Regression on Optimism (n = 2,488)
Model 1 Model 2 Model 3 Model 4
Origin x Marital Status -0.010 0.124
Origin x Children 0.081 0.062
Origin x Religion 0.024 0.089
Origin x Social Class -0.008 0.045
R2 0.024 0.081 0.082 0.083
F for change in R2 30.756 21.999 1.245 0.592
Dependent Variable: Optimism: Happiness and Self perceived health
*p < .05. **p < .01. ***p < 0.001.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
To conclude this section, Table 4-7 summarizes the results for each hypothesis tested.
Table 4-7 : Hypothesis Testing Summary
Hypotheses Test
Family dependence Partially proved
Attitudes toward family and elders Disproved Attitudes toward family roles Proved Attitudes toward women’s role in society Proved
Optimism Partially proved
‧
國立 政 治 大 學