Factor analysis

85

Table 20: Q85 You think micro-aggressions exist against foreigners in Japan.

Table 21: Q86 You have experienced micro-aggressions.

86

the tables always used the same items. The general life satisfaction (GLS) was compiled from 5 items: 1(low) to 10 (high) scale for life satisfaction was combined with 4 LIKERT scales and was tested with a reliability of .78 in Cronbach's α. The LIKERT items for GLS focused on being personally satisfied with their life in Japan. The factor used to represent life satisfaction professional (LSP) was meant to represent professional satisfaction and confidence of the individual, consists of 3 LIKERT items focusing on future economic stability, i.e. job security, room for professional advancement and whether or not they could find work if they happen to lose their current job or business. It scored a .66 on the Cronbach’s α, which is recognized as below the common acceptance of .70.

Two opposing composites recognizing cultural fit (CF) and frustration (FRUS) were compiled in factor analysis.¹⁹ They scored .78 and .84 respectively with Cronbach’s α. The CF factor consists of 5 LIKERT scale items. The 5 items included interaction with Japanese counterparts, having a support network, and feeling free to express him/herself to Japanese friends and/or co-workers. Finally, the ability to communicate within the community was also included, but this item did not specify which language they communicated in. The FRUS is made up of 9 LIKERT items dealing with the feelings of frustration, exclusion, stress, and feeling stuck in Japan and difficulties with cultural differences. The CF and CF FRUS

components act as opposition to each other and were expected to score that way from the start.

Both of these factors are also reflective of the connectivity and interactions by the individual with the host culture. This differs from the life satisfaction factor which primarily deals with the individual's satisfaction only.

During factor analysis two factors for language ability were recognized. The first was labeled language ability (LA) and consisted of 7 items. Four of those items were the self-rated 1 (low) to 10 (high) scales on Japanese speaking/listening/reading/writing skills. The remaining

19 This factor was labeled CF balance in O’Keefe 2016.

87

three were LIKERT scales items dealing with whether or not the individual feels confident in Japanese-only speaking situations or using the language in an emergency situation. The second factor covered daily language used by respondents (DL), which was compiled of three LIKERT items. The Cronbach’s α were .88 (LA) and .59 (DL). A subjective reason is all that can be given for the split between the two factors. The DL items may reflect daily language use which does not necessarily require high language ability. LA factor scored high on reliability and the weaker DL factor was used for reference to basic language needs.

Finally, the composites for discrimination were broken down in factor analysis into two components. They were labeled as discrimination and microaggressions with respective Cronbach’s α of .76 and .91. The 5 item component for discrimination is a compilation of LIKERT scales which include personal experiences with discrimination. The 2 LIKERT items for the microaggressions components represent whether or not respondents believe

microaggressions exist in Japan and/or if they experience them personally or not.

88 Table 22: Factor analysis 1

89 Table 23: Factor analysis 2

90 Table 24: Factor analysis 3

3.5.1 Hypotheses

Several hypotheses taken from a previous study were derived from the factors from the above section (O’Keefe 2016). The majority of them deal with the hybrid model’s styles of belonging. Life satisfaction and cultural fit act as the center lines for the typological results found in Chapter 5. The frustration factor used in the Pearson r correlation is not the same as the frustration type, but rather acts as the opposite to the cultural fit factor. Many of the items used for the frustration factor display conflict or difficulty with interaction with the Japanese community.

The first two hypotheses are related to common beliefs in the LTW group which often arose during the interview process. Language ability was tested with general and professional life satisfaction.

91

Hypothesis 1 a-b: General life satisfaction is positively correlated with a) language ability and b) daily language skills.

Hypothesis 2 a-b: Life satisfaction professional is positively correlated with a) language ability and b) daily language skills.

The next two hypotheses deal with the opposing nature of the cultural fit (CF) and the frustration (FRUS) factors. These factors are expected to act in complete contrast to each other.

The items for CF were focused on free expression of feelings in the community, the ease of communication, acceptance in the community and the observance of a support network. None of the items used for the CF factor asked which language they communicated in when interacting with the community. A respondent speaking only English or a combination of English and Japanese would be limited in some ways, but this may be the necessary form of communication for the community the operate in. The FRUS factor is constructed of items dealing with stress, feeling stuck in Japan and frustration in personal, professional and community-orientated relationships. Exclusion and feelings of cultural difference were also included. Language ability is not incorporated in the FRUS factor. Speaking the language, according to past research (O’Keefe 2013), does not mean one is void of culturally inhibited stress. There are several hypotheses that can be derived from this.

Hypothesis 3 a-c: Cultural Fit is positively correlated to a) general life satisfaction b) life satisfaction professional c) language ability.

Hypothesis 4 a-c: Frustration is negatively correlated to a) general life satisfaction b) life satisfaction professional c) language ability.

The factors for discrimination and microaggressions comprise the final pair of hypotheses.

It was a challenge during the interview process to get clear examples of discrimination due to the lack of strength in stories once certain bias are questioned. The separation of fact from subjective recounts of past incidents can prove to be somewhat problematic. This topic is often the subject of debate in online foreign groups, forums and editorials. The reverse side of this discussion is the overwhelming amount of positive discrimination many LTW respondents

92

received during their time in Japan which has been observed in past studies as well (Komisarof 2012). True social or institutional systematic discrimination would and should equally affect people in the same way. This particular study found little to no evidence of systematic discrimination, but could never claim that it doesn’t exist. On the other hand, examples of microaggressions were fairly common, but respondents were not equally affected. While microaggressions will be discussed more in Chapter 4, the hypotheses below hope to offer a quantitative view of the phenomenon.

Hypothesis 5 a-b: Cultural Fit is negatively correlated with a) discrimination b) microaggressions.

Hypothesis 6 a-b: Frustration is positively correlated with a) discrimination b) microaggressions.

3.5.2 Pearson’s r correlation

The attempt to measure their life satisfaction and its correlation to language ability and the cultural fit model (Ward and Leong, 2004; Chirkov et al, 2005) is to understand what factors determine success from the social and psychological angle. The factors for discrimination were also used in the bivariate correlation to get a multi-angled view of the situation. All eight factors were then tested for correlations using a Pearson’s r (Table 25) analysis and published with the mean values (𝜒) and standard deviations (SD). Due to the amount of items used for each composite the n varied for each correlation, which was also reported separately when necessary.

While this was fully reported in O’Keefe (2016), this section will report those findings again and reveal the relationships to this study. As explained earlier, this study is seeking to define various integrative habits of LTW from a social psychological manner over the long-term in Japan. There are some LTW use techniques to interact with the host culture while others separate themselves from it whenever possible. This study hopes to clarify these patterns through the forming of the typological groups.

93

The results of the Pearson’s r analysis observed the strongest correlation between the CF and LS factors. The 2016 (O’Keefe) study had several hypotheses, but this section will review 6 of the most significant findings. Cultural fit (CF) as explained earlier in this paper is the connection to the community. This connection does not necessarily represent the self-sameness experienced with the community needed for identity achievement, but it does serve as valuable information when determining the typological goals of this study. This is also information that has not been considered before about the western foreign group in Japan.

Table 25: Pearson correlation

One of the most significant discoveries in the analysis was the correlation between

language and life satisfaction. When respondents were asked the importance of language, it was common for them to put weight on its necessity. The result of the Pearson’s r was significant enough to say according to this survey, language ability does not have the strongest correlation to LS among the factors tested. There were two language factors which represented ability and daily use of language. The variance between the two factors representing language ability scored .30 for LA and .06 for DL when correlated to GLS.

This section will give a brief overview of the 6 results from the findings of O’Keefe’s (2016) paper. The first and second hypotheses were both tested with a sample of n = 286. The general life satisfaction composite showed a positive correlation with language ability (r = .30) but the daily language composite accepts the null hypothesis (r = .06), creating a split for the 1st

94

hypothesis. The 2nd hypothesis showed both language ability (r = .32) and daily language use (r

= .13, p ＜ 0.05) had positive correlations with professional confidence. The 3rd hypothesis can be reported also as displaying a positive correlation between CF and a) general life satisfaction (r = .73, n = 274), b) professional satisfaction (r = .35, n = 274) and c) language ability (r =.35, n

= 281). The FRUS composite for the 4th hypothesis can be recorded as negatively correlated with a) general life satisfaction (r = -.60, n = 274), b) professional satisfaction (r = -.31, n = 274) and c) language ability (r = -.22, n = 281). The 5th hypothesis (n = 233) can reject the null. The composite for CF communication was negatively correlated with both a) discrimination (r = -.41) and b) microaggressions (r = -.27). The 6th and final hypothesis shows that FRUS is positively correlated with a) discrimination (r = .59) and b) microaggressions (r = .39) respectively. All of the above correlations are significant at the p ＜ 0.01 level unless expressed otherwise.

3.5.3 Methodological limitations

In all, this study can only be considered exploratory because of its inherent limitations.

To begin with, although the respondents were from varying postal codes nationwide and were at least two separations from the researcher, the sample was not random and originated with the researcher on SNS networks. Moreover, even though the criteria for the study appeared at the beginning of the questionnaire and elicited approval from every respondent, this means it is completely reliant on the honesty of individuals who answered the questions. Through the process used it would be challenging to secure the accuracy of those who chose to answer the questionnaire. This could lead to the sample as not a true representation of the group. This type of personal bias has also been mentioned in other studies of ethnic groups in Australia (Mak &

Tran 2001). This bias can most often be noted in the language area. There may be some respondents who overrate their ability, while others may be more humble and underrate themselves. This was apparent in some of the preliminary testing, but to confirm language ability beyond the self-scored item is problematic. While those early preliminary questionnaires

95

have not been added to the final sample for this paper, they offered valuable information into how the survey in general was received. However, the self-rated scales could also be a

representation and reflection of how confident a person feels about their ability, which gives insight into their personal feelings of accomplishment and satisfaction with their language ability.

In hindsight there are items that may have performed better if worded slightly different.

This was brought to light by comments at the end of the survey. One example is how homeschooling was not offered as a choice. While not a common practice, it came into the discussion as a viable choice for some foreign parents. This was never covered in the survey.

Japanese would mostly likely never even entertain the concept of homeschooling. This may be especially true for families whose parents are both non-Japanese (O’Keefe 2013). Another topic brought to attention was the lack of choice for transgender individuals or couples. This will be considered in future work.