Preference of Norms Over Laws (by response)
4. Research Findings & Analysis
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
36
4. Research Findings & Analysis
In this chapter, relationships between the variables will be explored as well as the results and analysis of the data. The chapter is comprised of five parts: the first section will provide an overview of descriptive statistics which were used to analyze the demographic characteristics of respondents and responses to the items in the survey, the second section presents the test of reliability through Cronbach’s Alpha Value to confirm internal reliability of composite variables, the third section illustrates the correlational data conducted specifically through the bivariate correlation analysis (Pearson’s coefficient), the fourth section applies a linear regression analysis to the data to explore validity of this study’s framework , and the fifth and final section presents a summary of the findings and discussion of results.
Descriptive Statistics
This research consisted of conducting a survey on Taiwanese people from different age groups and backgrounds in order to understand their feelings and perceptions about migrant workers in Taiwan, social norms that generally oppress migrant
workers, the strength of confidence in rule of law, and migrant workers’ legal rights in Taiwan. An online questionnaire was used and participants were recruited through several methods: the researcher is part of a popular television show and so used this popularity to provide the survey to the television show’s audience, teachers of Erxin Primary School in Keelung, Taiwan participated in the survey, and through the researcher’s institution and peers the survey was also conducted. This purposeful sampling was used in order to ensure a wide variety of Taiwanese participants from different geographical areas and demographic backgrounds. There were 303 total participants, out of this number 296 were valid because some participants left items blank on the questionnaire.
The demographic characteristics were analyzed to provide a minimal
background of participants based on age, gender, and occupation. Below is a chart of these demographic variables:
Table 4.1 Demographic Characteristics of the Sample
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
37
Descriptive Statistics for Continuous Variables
In this section, mean scores, deviations and important trends will be discussed for the four main variables: mentality, social norms, rule of law, and marginalization. In this section, the items marked with an asterisk (*) are reverse coded items which have been positively coded and interpreted in reverse for the analysis. The higher the mean value for an item, the more strongly the participants felt about that specific attribute.
High standard deviations imply there was a large variety in responses to the items and low standard deviations would mean that most observations clustered around the mean value.
Mentality
The table below shows the mean values and standard deviations for the mentality variable. High scores (5) indicate agreement to the statements and low scores (1) show disagreement.
Table 4.2 Descriptive Statistics of the Mentality Variable
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
38
In this table, most respondents strongly agreed with the launch of the new
governmental smartphone application that allows any citizen to scan a foreigner’s identification card as this item had the highest mean score: 3.64. The item that had the lowest mean score shows that most people disagree with the statement that migrant workers should not have the same labor rights as Taiwanese workers. Below is a graph summarizing the mentality responses by percentages and there are six items dealing specifically with people’s feelings towards Americans and migrant workers.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
39 Figure 4.1 Responses to Mentality Variable items
Within this mentality section, a portion of the items was dedicated to explore the difference in feelings towards Americans working in Taiwan and migrant workers working in Taiwan in order to find out if disparity existed. According to the responses seen above, 56% of participants felt that migrant workers should be allowed to
become citizens (X.31.ME) and 68% of participants also felt that Americans should be allowed the same right (S.10.ME). 55% of participants do not feel that migrant workers take jobs away from Taiwanese people (X.25.ME) and the same percentage is found in regards to how the participants feel towards Americans (X.29.ME). Lastly, 56% of participants feel that migrant workers should be punished if they run away from their employment (X.9.ME) and only 7% felt that Americans should be punished if they run away from their employment (X.16.ME). These figures potentially indicate that there is a greater tolerance or acceptance of Americans as opposed to migrant workers in Taiwan.
Social Norms
Below is a table detailing the means and standard deviations for all items pertaining to the social norms variable.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
40
Table 4.3 Descriptive Statistics of the Social Norms Variable
In this section, common social norms associated with employers and migrant workers that were found during the qualitative research of this study are explored. The highest mean score of 3.1 is seen in agreeance with the requirement of caretakers living with the employer if the employer deems it necessary. Most items in this section have a low mean score which indicates respondents in fact answered in a way that disagreed with many of the norms that are in practice today. For example, below is a chart showing that 95% (X.24.N) of participants feel that an employer must adhere to the labor contract and not be allowed to perform illegal punishment, and 93% (X.19.N) feel that an employer should not be allowed to force the worker to conduct other responsibilities not listed on the contract (which all of these were found in the qualitative analysis to be widely accepted). This disparity between findings in qualitative research and this survey could be explained several ways: respondents could have been reluctant to answer in a way that would cast them in an unfavorable light or there could have been issues in the wording of the items.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
41
Figure 4.2 Responses to Social Norms Variable Items Marginalization
The following table displays the items belonging to the marginalization variable in which migrant workers are placed in a more vulnerable position to exploitation. All items for this section were reverse coded.
Table 4.4 Descriptive Statistics of the Marginalization Variable
Statement MA28 holds the highest mean score of 2.82 showing that respondents erred on the side disagreement in regards to allowing migrant workers the right to vote for
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
42
pertinent legislation. The second highest score of 2.32 indicates respondents felt uneasy or undecided about allowing migrant workers to rent their own housing independently of their work. These means show that most of the respondents answered in favor of improving the status of migrants workers, i.e. letting them unionize and allowing them more job mobility. Once again, this trend could be contributed to several factors: almost half of the sample were younger in age which could have effected their outlook on migrant workers or many could just be
uncomfortable with answering too honestly.
Rule of Law
The table below details items pertaining to the rule of law variable in regards to judicial systems and government involvement. All items were also reverse coded.
Table 4.5 Descriptive Statistics of the Rule of Law Items
The highest mean score of 3.77 shows that respondents do not strongly believe or trust in the judicial system of Taiwan and the second highest mean shows that a portion of the respondents do not feel the courts are unbiased in their legal processes (mean 2.83). The lowest mean score of 1.98 shows that most participants do feel that the government is more important than the community in maintaining social order.
Therefore it is found that there is a positive affinity for a strong government, but that there is also a lack of trust in the judicial system of Taiwan. If citizens do not trust their government, for example in settling of legal disputes as surveyed above, it is reasonable to assume that people then are less likely to desire governmental
involvement in their personal affairs. Without strong governmental regulation of the treatment of migrant workers, people are left with less incentive to treat their workers fairly.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
43
Descriptive Statistics for Demographic Variables
After compiling the descriptive data, intriguing trends were found for both the gender and age demographic variables. Occupation did not seem to follow any
particular pattern or show any trends and therefore will not be analyze din detail.
Below are four figures placed in a grid to display age groups and response mean scores:
Figure 4.3 Summary of Mean Scores of Continuous Variables Stratified by Age In the figure above, several noteworthy trends appear. Firstly, in the top left it shows that older respondents responded negatively to statements regarding marginalization.
In this figure, the items are not reverse coded which means that as age increased, more and more respondents felt that migrant workers should not have certain legal rights that would improve marginalization conditions such as job mobility,
accommodation rights, or voting rights. Items for the rule of law variable were not reverse coded for this figure as well, and shows an opposite phenomenon: as age increases, so do people’s trust and belief in the rule of law within Taiwan. The mentality and social norms variables on the right side of the figure show the general
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
44
trend that as age increases, respondents answered more discriminatory towards migrant workers, and more in favor of social norms that in fact perpetuate migrant worker abuse. Overall, those respondents that were older answered in favor of more control over migrant workers; this could be attributed to socio-historical factors that come with living in a strongly homogenous society with protective mindsets.
Below is a similar figure displaying the mean scores stratified now by gender:
Figure 4.4 Summary of Mean Scores of Continuous Variables Stratified by Gender In the figure above, it is evident that the different genders generally solicit different responses depending upon the continuous variable. For the marginalization and rule of law variables on the left side of the figure, the data shows that males generally are not as supportive as females in reducing marginalization and also males have a much higher trust in the rule of law in Taiwan. On the right side of the figure, males again answered more discriminatorily and more in favor of social norms than women respondents. This has significant implications because of the strong patriarchal nature of older Taiwanese society; in other words, it is often the man of the family who hires the migrant worker, and factory/fishery employers are most often male than female.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
45
Paired T Tests for Gender Influences on Feelings towards Americans and Migrant Workers
In the descriptive statistics, the general trend was found that males answered more discriminatorily than females, and so in this section that trend will be explored further to analyze feelings towards migrant workers and American workers in Taiwan by both the male and female respondents. Out of the six items on the survey that measured feelings on this subject, only one comparison was found to have a large difference in responses: 56% of responses believed that migrant workers should be punished if they quit their job, while only 7% thought Americans should be punished.
Therefore, it is hypothesized that there is a difference in responses between these two categories, and that this difference is positive meaning respondents who feel
discriminatorily towards migrant workers do not feel discriminatorily towards American workers.
The null hypothesis would be the difference in means is zero or less than zero meaning respondents felt the same way about migrants and Americans or they are more discriminatory towards Americans than migrants. Below are two Paired T Tests which analyze males’ responses and females’ responses to the two items.
Table 4.6 Male Paired Sample T Test
As found in the table, the T(119)=13.763, p<.001, which means the null hypothesis is rejected. Males did answer discriminatorily towards migrant workers, and not so towards American workers.
Table 4.7 Female Paired Sample T Test
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
46
In the female table, T(174)=11.907, p<.001, which means females also responded more discriminatorily towards migrant workers and less so towards American workers. However, when the two tests are compared, it is evident that the mean responses are much higher for males (1.591) than for females (1.257), showing that as a whole, the male respondents were more discriminatory than the females.
Reliability Analysis
The reliability test used in this research is Cronbach’s alpha coefficient test which analyzes the internal consistency of the scales used in the survey. Below are the four main variables with their subsequent Cronbach alpha coefficients:
Table 4.8 Cronbach’s Alpha of the Measures
In the table above, all variables are above .6 and are therefore acceptable. The social norms variable has the highest Cronbach alpha value of .72, with social norms as second with .7.
Correlations and Linear Regression Models Correlations between Variables
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
47
In this section, major relationships between mentality, social norms, rule of law, and marginalization are discussed using a visualization where histograms, scatterplots, and bivariate correlations of the four outcomes are displayed on diagonal blocks. In the figure below, blue lines represent normal distributions with means and standard deviations equal to the sample means and standard deviations of the respective averaged Likert score. On the lower left blocks, scatter plots of the outcomes are displayed with red lines indicating general trends. On the upper right blocks, the absolute value of the sample correlation between the diagonal outcomes is printed with font size proportional to the degree of correlation.
Note: **=p<.01
*=p<.05
Figure 4.5 Visualization of correlations between outcomes
**
* **
.22*
.43***
.19**
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
48
According to the figure, the strongest overall correlation is noted between mentality and social norms (r=.48, p<.01), where the R2 value of 23% of the variance in social norms outcomes is explained by mentality scores. This correlation supports H1 and the second highest correlation is between marginalization and social norms (r=.45, p<.01), where the R2 value of 20% of the variance in social norms is explained by marginalization which supports H4. There is also a highly significant relationship between marginalization and mentality (r=.43, p<.01) which also supports H4. The relationships between rule of law and mentality and rule of law and social norms are lower and less significant, but still noteworthy. There is no correlational relationship found between rule of law and marginalization.
Linear Regression Analysis
In the overall framework, it is hypothesized that mentality towards migrant workers, acceptance of social norms that abuse migrant workers, and the distrust of rule of law all contribute to the marginalization of the migrant labor population. This linear regression model below explores the correlations between the measures and tests whether these predictors are significant in influencing the final outcome of marginalization.
Table 4.9 Linear Regression Analysis of Mentality, Social Norms, and Rule of Law on Marginalization
Several important data trends can be found in the table above. Firstly, it is found that about 25% of the variance in marginalization responses can be accounted for with the three predictors (r2=.252, p<.001) of mentality, social norms, and rule of law.
Secondly, the t-tests for the beta coefficients for mentality and social norms as well as the F Value of the model are all highly significant. With these large values, it is
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
49
evident that the model’s regression is big enough to have significant predictive capabilities. Therefore, the framework of this research is validated.
Summary and Discussion of the Results
This section will discuss the summary of the data in relation to the hypotheses of this study. The table below shows that H1 and H4 are fully accepted, H2 is rejected, and H3 is partially accepted based on the analysis of the results. Further relationships and significance of these relationships are discussed below.
Table 4.9.1 Hypotheses summary