Attitudes to Genetic Engineering

Japanese and International Comparisons

Darryl R. J. Macer, Ph.D. Eubios Ethics Institute 1992


Copyright 1992, Darryl R. J. Macer. All commercial rights reserved. This publication may be reproduced for limited educational or academic use, however please enquire with the author.

2. Samples and Method


page 10-18 in Attitudes to Genetic Engineering: Japanese and International Comparisons D.R.J. Macer (Eubios Ethics Institute, 1992).


There are limitations of opinion poll surveys. They can only provide broad indications of attitudes and perceptions, and opinions are subject to change. Nevertheless, information on the structure and underlying determinants for attitudes may be revealed, and this method is the most efficient and practical for revealing these.


2.1. Sample Selection

There are some cultural difficulties which are important when conducting a public opinion survey in Japan. People are not used to having people, except for religious groups and sales-people, come to their house door, and ask for an interview. They are also not very receptive to on-the-street questions, and in light of the relatively long time (20 minutes) that is required to complete this questionnaire, this method was also not very successful. Although some short interviews for opinion polls are conducted with a relatively low refusal rate, these include the gift of a present, such as a telephone card, to the subjects. Therefore, this approach was not possible for two reasons, one being that the questionnaire was relatively long, another being the shortage of funds. An attempt to obtain a Japan-wide sample was made using students as interviewers, to conduct the questionnaires during their 1991 summer holiday travelling throughout Japan. However, this could not provide a completely random sample because of a high refusal rate by the public, and the use of relatives and friends, and the limited number of localities that could be tested using this approach. It did, however, provide a useful introductory trial, and some useful opinions were expressed in the answers.

A mail method shares the disadvantage that a statistically random group of respondents is not chosen. Nevertheless, in view of a high refusal rate from household or on-the-street interviews, it was still chosen as a method to provide a larger sample population. Japanese people may also feel easier completing such a survey in the absence of other people watching them, and may feel freer to express their opinions. In October 1991, 1500 questionnaires were delivered anonymously, with a further 200 questionnaires delivered by students to members of the public. The interview responses were similar to those obtained by the mailed questionnaire, except that the respondents to the mailed questionnaire wrote lengthier comments.

All surveys need to make samples, to reduce the time and costs of a survey. The total population cannot be surveyed, so samples were chosen. These samples include groups that may be influential in future policy discussion of these issues. Sampling can be almost as accurate as studying the whole population depending on the questions that are used and the analysis of the responses (Guy et al. 1987). The types of sampling method include simple random systematic samples, which were used for the high schools and University staff. From the name lists, addresses were chosen using names spaced at regular intervals. The public sample was sampled in the following way; households in rural and urban areas, and in different prefectures were randomly selected, from particular clusters of households that were chosen for random mailing. One variable is the presence of a letter box, not all households possess a letter box, especially they may be absent from farms, so that considerable time was spent in distribution. Samples of detached houses and apartment blocks in urban areas, and farms were chosen. The samples were chosen based on both convenience and representativeness. It was not possible financially to sample every district of Japan. Questionnaires were distributed by hand by 8 people in different areas of Japan.

The method used to obtain opinion from the specific groups, high school biology teachers, scientists, and Tsukuba University staff, was also a mail out questionnaire. The selection method used for samples varied.

The complete list of Japanese high schools was used to randomly select a sample of 502 high schools representing every prefecture. A letter was sent to the headmaster of each school explaining that it was requested that they chose a biology teacher (if not available, than a science teacher), to give the questionnaire to. The selection method of each teacher by the headmaster was not specified, because it was considered inappropriate to constrain the headmaster by such a method. This is not considered to introduce a significant bias in the questioning procedure because the response rate expected was less than 50%. The English translation of the letter sent to headmaster's is below, it includes a traditional summer greeting common for formal Japanese letters.

University of Tsukuba, Institute of Biological Sciences
July 1991
Dear Headmasters,
I hope all the teachers are fine in the heat of summer. Please forgive the sudden request for answering a questionnaire.

This survey is planned by the group of researchers in the University of Tsukuba, in order to discover the public attitudes toward science and technology. The survey has been conducted through an internal project of the University, using funds from the Ministry of Education. We would be grateful if you could give this questionnaire to your biology (or science) teacher as we need their answers.

We hope for the further development of your school.
Sincerely,
Darryl Macer

The sample of University staff was chosen by selecting one name from every four names in lists of the University staff. The scientists group consisted of three subgroups, and the names were randomly chosen from lists of scientists that were published. There was not complete list of scientists of all ages in Japan, so samples were taken from several lists. These included a list of people belonging to particular scientific committees, and lists of staff at scientific and academic research institutions throughout Japan, and in Tsukuba.

Tsukuba University students were sampled by selecting several classes and student clubs at random, and asking members to respond (a total of 300 were asked).


2.2. Questionnaire Design

The questionnaires were written on three double-sided sheets of A4 paper stapled together, with the introductory note on the first page of the questionnaire. They were sent with an enclosed stamped and addressed return envelope for return to a P.O. Box number. University of Tsukuba envelopes were used for the mailing.

The questions were chosen not to be leading, so as not to sensitize people to the issues that were addressed. The order of the questions was also important, so that questions about specific developments, or applications, such as genetic screening, gene therapy, or the production of particular genetically modified organisms (GMOs), were given after general questions. A copy of the questionnaires is in the appendix, chapter 10. The questionnaire title was "Science and Technology survey", so not to sensitize people to the fact that it was specifically concerned with biotechnology and genetic engineering, though after completing it some people may have realised that. The items chosen for the questions were considered appropriate for the purposes of the project, and from the results other questions can be developed for future surveys that may elucidate certain aspects of people's feelings that were not revealed in these questions.

An introductory note was given on the questionnaires, which is shown below. The note for the scientists and teachers (the note for the University staff was essentially the same) was:

This is a survey being conducted by international researchers at Tsukuba University. It is being conducted under funding from the Ministry of Education. We would like to find out what you think about science and technology. We want to ask you some questions. This is a confidential survey, we want you to express any opinions that you wish.

You were randomly selected, and we appreciate your cooperation, freely sharing your precious time. The number written on this form is for verification purposes only. It can be completed in 15-20 minutes, but you are encouraged to write more if you wish. The answer can be given in Japanese or English.

The note included for the public (the student note was similar) was:

This is a survey being conducted by international researchers at Tsukuba University. It is being conducted under funding from the Ministry of Education. Please excuse me, we have randomly chosen you, we will ensure that the result will be confidential, so please give us your candid opinion on science and technology. The results will be published shortly. We would like to find out what you think about science and technology.
Please hand this questionnaire to the member of your family over 15 years old whose birthday is next, and who wants to fill it in. It can be completed in 15-20 minutes, but you are encouraged to write more if you wish.
Even if there are some questions which you do not understand, it is useful to have your answers, so please attempt to answer all questions. We cannot give you anything in return for your time, but we hope that this survey will have some influence on the behaviour of scientists and the government.

2.3. Sample Response

A certain level of non-response in surveys is constantly faced. In interviews, people may refuse to talk, and in mail out questionnaires many do not reply. The level of response will reflect several factors, such as the interest of the people in the contents of the questionnaire, the length of time needed to complete the questions, the complexity of the questions, and the attitudes to the organisation performing the research. Sample surveys are a major method of social science research, and there are different purposes. One purpose of the current survey was to obtain quantitative data on the attitudes to particular questions. The level of uncertainty of the results depends on the response rate. Another purpose of the study was to obtain ideas of how people feel about these topics, and concerns that they have. This part of the survey is less affected by the response rate, because those individuals that have strong feelings are likely to respond to the questionnaire, and the range of negative and positive feelings should be able to be determined.

Some people do not like to give information about their views to others, even if the survey is anonymous. However, a more common reason for nonresponse is that people are busy and do not want to spend the time to complete the questionnaire. There is a danger that this study will overlook the opinions of the silent majority who did not respond to the survey. There are considerable differences between countries, in Western industrialised countries a response rate of 65% is commonly accepted as a benchmark (Goyder 1987), however, in Japan a response rate of 30% is common. The response rate obtained from scientists in Japan and New Zealand was 56%, and 58%, respectively. These were similar despite the use of a lottery charity prize in New Zealand and the absence of one in Japan. The high response rate could reflect a high level of interest in answering the questionnaire in Japan by the scientists, compared to other groups. The response rates of the biology teachers were 45% in Japan and 64% in New Zealand. The difference could be due to the use of a lottery school prize in New Zealand, whereas there was no incentive or reminder letter used in Japan. However, because the response rates of scientists in both countries were similar, it may be due to other factors. The response rate from the University of Tsukuba staff was 35%, which is disappointingly low, but may reflect the high exposure of this group to questionnaires.

The response rate of the public to interviews cannot be determined, but a high rejection rate was experienced. This led to the use of a mail out questionnaire, from which a response rate of 26% was obtained, which is satisfactory for Japan. The response rates obtained in other opinion surveys on the life sciences and bioethics vary. The response rate of the face-to-face interviews of the Prime Minister's Office (PMO) have been about 74% (N=10000, Dec. 1985, and N=3000, Oct. 1990), and of Dentsu research survey on biotechnology in Sept. 1985 a response rate of 69% (N=1400) was obtained. The response rate to the mailed survey of readers of Nikkei Business in July 1983 was 40% (N=1500). However, the response rate of a mailed survey of readers of the science magazine Newton (1989) was 60% (N=500). Recently, there have been more frequent opinion surveys in Japan. The response rate of face-to-face interviews over mail interviews in Western industrialised countries is about 8-10% greater in this range (Goyder 1987), but in Japan the interview surveys achieved a 40% higher response rate, though the groups are different and cannot be directly compared. In the current survey, the mail-questionnaires obtained a better response rate than the direct door knocking-style interviews or on-the-street interviews that were attempted. A February 1991 mail survey of selected "public monitors" by the Agency for the Environment (EA 1992) had an extraordinary response rate of 91%, but this was not a random sample. Rather, the questionnaire was sent to a sample of the population who had already agreed to be monitor survey respondents.

People living at the margins of conventional life are difficult to survey. One method used to determine the attitudes of a wider range of people is to use selected sampling techniques, and different methods. Unless the pool of nonresponders represents a perfect subfraction of the population, nonresponse will bias the results. Methods used to promote response include mailing of reminders, however, this requires additional expenditure. In the case of the public mail out survey, this would not have been possible in this study, without considerably greater cost because addresses were not recorded (to save the cost of stamps and the labour costs of addressing envelopes), rather the questionnaires were randomly dropped into letterboxes of people living in different regions of Japan. There may be less bias in the joint results from mail and face-to-face interviews.

Involvement with the topic may be one positive factor in response, as seen from the higher response rates of scientists compared to high school biology teachers and the public in general. However, comparisons of the response rate from members of different University of Tsukuba Departments were made, and these actually revealed that administration personnel had the highest response rate within the University, and teaching staff had a lower response rate. The responses to answering different questions on the survey depended on the location in the questionnaire, with some respondents missing out the penultimate page of questions, whereas almost all answered the last page which included personal details. Perhaps they became tired of the lengthy questionnaire, because in some samples the position of the questions was altered, but there were still respondents who did not answer the penultimate page but answered the final page regardless of the questions on it. There is no easy relationship between response and involvement, though it is probably related at least to some degree to understanding and/or interest in the questionnaire.


Table 2-1 Cumulative response rates of the different groups

Cumulative response rates, expressed as %'s of the questionnaires that were mailed. The number delivered is the total number of questionnaires that were mailed minus those returned because delivery was not possible.

Sample:
Public

(mailed)
HighSchool Biology Teachers
Total

Academics
University of Tsukuba

staff
Outside University of Tsukuba
Number delivered1500 5021514 665849
Total Number returned 385228 728249 478
% Response Rate25.7 45.448.1 37.456.4
Week of Return (Cumulative response rates, %'s)
113.3 12.417.8 15.819.4
220.9 29.137.8 31.343.3
323.1 35.943.7 34.751.1
424.6 38.445.1 36.152.6
524.8 39.845.8 36.853.4
625.4 40.846.8 37.054.8
725.5 41.647.3 37.355.3
825.6 42.647.3 37.355.3
925.7 43.447.6 37.455.9
10+ 45.448.1 56.4

There are various types of potential error. Systematic error could be induced by cues from the interviewer or on the questionnaire. The specific topic of the questionnaire, genetic engineering, was not stated, so as not to bias the respondents from the start of the survey, especially when answering question Q5. Respondents may also select the response that they think is the preferred response for the researcher, so the order of response of some questions was altered, from approve to disapprove, or benefit to no benefit, to attempt to avoid setting a pattern of preferred responses. Random errors principally mean the errors of using a nonrandom and imperfect sample for modelling the population. Random errors also include some mistakes on the answer forms that people may make, for example in Q16, a few people could put answer 1 instead of 5, if they made a mistake in reading the questionnaire. Some check was made on the consistency of the answers, but this could only be done with certainty in the responses to questions where the respondent also wrote a comment explaining their feelings. Measurement error includes some instability in the opinion of people, for example, if they read a particularly impressive news article the morning before they answered the questionnaire. The values expressed in tables include tenths of one percent, but this does not represent the reliability of the responses as estimates of the population statistics. The estimated sample errors for the questions are several % depending on the number of respondents and the response rates.

The questionnaires to special groups, except those to the public and students, were numbered in order to record the reply time and to determine the response rate from particular groups, especially if respondents did not state their address or employer. Numbering allowed comparison of the replies from different localities and from private versus government employees in the case of scientists. The issue of personal privacy arises with the use of numbering. There was surprising low expressed concern about the presence of numbers on the questionnaires, which will be discussed later. Perhaps people trusted that the introduction ensured them that confidentiality would be maintained, or else some did not return the questionnaires because of privacy concerns. One out of 800 scientists, and 2 out of 700 staff, returned blank questionnaires stating that they would not complete the questionnaire because of doubts that their privacy would be maintained. 12 scientists and 3 of the University staff returned questionnaires with the numbers cut off.

Some comments were made by non University of Tsukuba academic respondents on the absence of a personal name in the introductory comment on the questionnaires. A few said that it was rude not to give the individual's name, therefore in the public questionnaire my name was included in the introductory note, as it had been with the school teachers and University staff. It is unknown whether a foreign name on the questionnaire affects the response rate.


2.4. Sample Characteristics

General statistical information was gathered concerning the sex, age, marital status, number of children, occupation, education, income, and locality of the address. The questions asked include questions on the sex, age (years), occupation, speciality of the academics, marital status(not married, married, divorce/widowed, other), children (none, pregnant, some (number )), education (High School, two year college, four year University, Masters, Doctorate, or other), monthly family income (categories: 1 The overall features of the samples are summarised in Table 2-2. The academic samples are distinguished by the huge majority of male respondents over female respondents, which represents the populations chosen, and the overall situation in Japan. The public and student samples were more balanced, in terms of sex. The public sample includes all those respondents from the public interviews (N=94), mailed out questionnaires (N=385), and questionnaires passed to the public by contacts (N=72). The university student sample includes students in the specific University of Tsukuba sample (N=151), and university students among the public respondents (N=53). The academics includes respondents both from the University of Tsukuba staff (N=249) and from other academics groups (N=479). The "total scientist" figure includes those respondents from the total academics who specialised in the natural sciences (including 113 University of Tsukuba staff, and 442 respondents from the academics outside the University of Tsukuba), defined as those who did not have a speciality in social sciences, economics or law, or in administration (see Table 2-4 for the academic specialities).

The age distribution of all samples, except the students, included people of all ages (the range of the public sample was from 15 to 90 years of age). The University student age in Japan tends to be exclusively young, reflected in the absence of any students older than 27 years in the sample. The average income of the academics is higher than the public sample, especially the average of the scientist's group, and 64% of the privately employed scientists had a family income of more than a million Yen a month.

An approximate estimate of the extent of national coverage of the sample can be obtained from the addresses. Public respondents from 20 of the 47 prefectures of Japan were included, student respondents from 37 prefectures replied, the academic sample included respondents from 25 prefectures, and high school teachers in every prefecture of Japan responded. The occupations of the public respondents are summarised in Table 2-3, and the specialities of the academics in Table 2-4.

Of the respondents in the so-called academic sample, 24% were private company employees, 31% were government employees (non-University), and 45% were University employees (almost all being at national Universities). The respondents in the scientist sample included 28% private company employees, 39% government employees, and 33% University employees. The work types of the University of Tsukuba respondents included 45% administrators, 23% lecturers and associate professors, 14% professors, 12% medical staff, and another 5% involved in research or who were technicians.

For teachers, information about the school was also asked:

What is the number of students at your school?
1 0-200 2 201-600 3 601-1500 4 1501+
What type of school do you teach at?
1 Public high school in urban area 2 Public high school in rural area
3 Private high school in urban area 4 Private high school in rural area

81% of the responding teachers were from government-funded schools. Of the total public and private schools, 66% were urban schools and 34% of the schools were in rural boroughs. Only 5% of the schools had less than 200 students, 16% had between 201-600 students, 68% had between 601-1500 students, and 11% had more than 1501 students.


Table 2-2 Sample Characteristics (%'s)

Sample:
Public
Students
High School Biology Teachers
Total Academic Sample
University of Tsukuba Staff
Outside University of Tsukuba
Total Scientists
Number551 204228 728249 479555
Male52.9 53.090.0 88.877.8 94.690.2
Female47.1 47.010.0 11.122.2 5.49.8
Age
>204.6 28.60 0.20.4 00
20-2923.7 71.417.1 9.615.0 6.99.3
30-3922.8 031.6 21.728.9 18.018.3
40-4925.2 021.0 31.228.9 32.231.5
50-5911.9 026.3 28.321.1 32.230.5
60-698.7 04.0 8.55.7 10.09.8
70+3.1 00 0.50 0.70.6
Marital
Single29.0 98.523.3 14.524.6 9.312.4
Married66.3 1.575.3 84.074.6 88.886.1
Div/Wid4.3 01.4 1.40.8 1.71.3
Other0.4 00 0.10 0.20.2
Children
None34.7 10029.0 19.729.0 15.017.7
Pregnant1.9 02.3 0.90.9 0.91.0
One14.1 018.9 14.416.2 13.412.8
Two32.2 033.7 46.340.2 49.446.8
More17.1 016.1 18.713.7 21.321.7
Education
High school36.6 58.40 10.425.9 2.43.4
2yr.

college

22.1 26.70.9 6.713.3 3.35.8
4yr.

college

30.9 11.980.4 36.323.3 43.138.0
Masters5.1 3.016.4 18.49.6 22.921.0
Doctorate1.7 01.4 25.525.4 25.528.4
Other3.6 00.9 2.72.5 2.83.4
Family monthly income
<¥500,00030.5 30.826.7 19.029.2 13.815.9
<¥800,00033.3 27.442.0 26.735.2 22.323.6
<¥1,000,000 17.826.0 20.722.5 17.824.7 24.0
¥1,000,000+18.4 15.810.6 31.817.8 39.236.5


Table 2-3 Public Occupations

The total number of individuals and % of total is recorded on the table.

Occupation
N
%
University student 5311.3
High school student 183.8
Retired34 7.3
Housewife108 23.0
School teacher12 2.6
University Staff or Researcher 122.6
Public Servant38 8.1
Company employee118 25.2
Office worker11 2.4
Business, Service Industry, or Self-employed 224.9
Engineer or other professional 71.5
Pressure group member 40.9
Arts5 1.1
Farming6 1.3
Not in full-time work 51.1


Table 2-4 Academic Speciality

The specialities of the academic samples. The total number of individuals and % of total is recorded on the table.

Sample:
Total
University of Tsukuba staff
Outside of University of Tsukuba
Scientists
SpecialityN %N %N %N %
Biotechnology, Applied Genetics 7410.23 1.27114.9 7413.3
Biology & Biochemistry 11415.816 6.59820.6 11420.5
Food Science14 1.900 142.914 2.5
Chemistry & Chem.Engineering 7410.23 1.27114.9 7413.3
Physics & Mathematics 415.616 6.5255.2 417.4
Administration & Econ. & Art 13618.8112 45.3245.0 00
Social Sciences & Law 334.622 8.9112.3 00
Medical92 12.75723.1 357.392 16.6
Earth & Ocean Science 172.43 1.2143.3 173.1
Engineering63 8.762.4 5712.063 11.4
Computer Science21 2.952.0 163.421 3.8
Agriculture45 6.241.6 418.645 8.1


Please send comments to Email < asianbioethics@yahoo.co.nz >.

To Attitudes to Genetic Engineering Contents
To Attitudes to Genetic Engineering Chapter 3
To Eubios Ethics Institute home page