With this research proposal, the largest difficulty was finding the subjects and getting people to agree to spend nearly an hour talking about their experience with the healthcare system. Within Taiwan, finding the people for Group A was not difficult but it was time consuming. Arranging a meeting and conducting the interviews was fairly straightforward, however, I severely underestimated the amount of time it took to transcribe and code the interviews.
Group A has A4.1 and A4.2 as subject designations because these two people were a married couple who were interviewed at the same time, but gave separate answers.
Group B was recruited in a similar way, and subjects in Taiwan were kind
54
enough to put me in touch with relatives in America. Many of the people declined an interview, but were kind enough to fill out the survey and return it. Group B had five in-person interviews along with six returned surveys. One of those surveys (B6) was disqualified because the person returning it was the American husband of one of the subjects. It was absolutely imperative that every one of the subjects of this research have experience of living in Taiwan under the NHI after 1995. If they hadn't done so, then their ability to evaluate the NHI became meaningless for the wider aspects of public policy because they won't have been subject to the same perception influences that other Taiwanese citizens have. For group C, the difficulty was finding subjects to email who would be willing to take the time for an interview, but there were five experts who kindly consented to sit down for an in depth interview. However, only four of the five health experts filled out the Scaled Data part of the survey.
Of the twenty-two interviews, nine were conducted entirely or primarily in Chinese. The rest were in English. After the interviews were finished, they were transcribed and coded. For the Chinese interviews, this involved an initial draft with consultation with a Taiwanese person for words and phrases that were difficult to understand. All of the data on laypeople was anonymized before being shown to anyone else.
As stated earlier, the shift from four groups to three groups was accomplished by not splitting the healthcare experts between those who had studied in the United States and those who had not. This represented an implicit assumption that the independent variable of having expert knowledge was a more powerful predictor of perception than experience living abroad. This also represented a practical consideration as four out of the five health experts had studied in the United States. This seemed to be the case with Group C giving remarkably similar answers on a range of issues, but with five respondents, such a conclusion cannot be drawn definitively.
Even though this was a small sample size, it was important to insure diversity within it. I made sure that there was variation among the age, profession and gender of those being interviewed. That said, while an effort was made to get as broad a sample as possible, the final groups were not representative of Taiwan's population. As a whole, the sample of laypeople was very educated. Among Taiwan's population, only 39% of the
55
population has a bachelor's degree or above84 while eleven out of the seventeen layperson subjects had graduate degrees. That's 64% of the sample when the actual population proportion in Taiwan is 4%85 so this group is not as generalizable to the population at large. However, the fact that both groups were both highly educated makes it easier to compare them to each other. The age distribution of both groups of laypeople was also fairly similar. It seems representative of people who are professionals in the middle of their careers. Income wise, the groups were somewhat comparable. They were also wealthier than the population at large.
The interview format evolved and changed as more people were interviewed.
Balancing the need for consistency vs. gradual refinement was somewhat difficult. In the Variables section, I detailed how some of the questions changed over time. There was also a gradual problem of formatting the surveys themselves. Due to word processing issues, some of the numbering on the questions was not consistent from one person to the next. They were all asked with the same wording and the same order, but a question might have been 2A) for one respondent only to be 2.1 on a later version. All of this was corrected when the responses were coded with each question getting a consistent
designation.
There was also an opportunity for an unfortunate kind of bias arising out of using people who had been college age while they were in the US. Foreign students in
American universities are generally insured through their institutions if they don't have direct insurance through their parents. This means that one aspect of the negative experience of living in a non-universal health system might not be obvious to those students since they aren't responsible for the cost of their own care. As it was, two people (B1,B2) in Group B had Student insurance while another two people (B5,B8) had
parental insurance. Three people (B7, B9, B11) had employer provided insurance in the United States and one person (B3) couldn't remember. Further complicating the data from Group B is that two subjects (B4, B8) never used the health system while they were in the United States.
The variance within the sample and the small sample size means that addressing
84 (“39 Percent of Taiwanese Hold Degrees in Higher Education” 2010)
85 (Hsiou and Pylypchuk 2012 p. 784)
56
the research questions relies on going through the individual subjects. The important point is to see the mechanism of living in the United States changing the perception of individual Taiwanese people.