Questionnaire development for the public opinion survey began with an examination of previous surveys conducted in Alberta and elsewhere in Canada on the topics of immigration, multiculturalism, and related themes. A few of these questions were included in the draft questionnaire. A larger number of original questions were devised by the research team. In addition, some of the questions asked of refugees themselves were included in the public opinion survey (modified, if necessary, for the different population being surveyed).6
The questionnaire and interviewing protocols were examined and approved by a University of Alberta Research Ethics Committee before interviewing began. A pretest with 30 randomly selected subjects led to the modification of some questions, and the decision to drop some others since the pretest interviews were somewhat longer than desirable. The final interviews averaged 15 minutes in length. Volume 3, Part II contains a copy of the final draft of the public opinion survey questionnaire.
The seven host communities for refugees destined to Alberta (Calgary, Edmonton, Fort McMurray, Grande Prairie, Lethbridge, Medicine Hat and Red Deer) formed the primary sampling areas for the public opinion survey. The target population in these communities consisted of all persons 18 years of age or older who, at the time of the survey, were living in a non-institutional dwelling unit (e.g., private dwellings in contrast to nursing homes or prisons) and could be contacted by telephone. The goal was to interview 150 randomly selected adults in both Calgary and Edmonton, along with 100 respondents in each of the five other host communities.
A random-digit dialing approach was used to ensure that all potential respondents had an equal chance of being contacted, whether or not their household was listed in a telephone directory. The Population Research Laboratory (PRL) uses a database of five-digit telephone number banks (i.e., 403-xxx-xx) covering all of Alberta. The seven sub-samples were drawn from the telephone database by using a computer program to select (with replacement) a simple random sample of banks for each community. The computer than appended a random number between 00 and 99 to each number selected. All duplicate numbers were eliminated from this randomly-generated list. Once the telephone survey began, business numbers and other non-eligible numbers (46% of all numbers randomly selected) were also eliminated from the sample, once they were identified. When contact with an eligible household was made, a quota sampling technique was used to select male or female respondents age 18 and older.
Table 2-4 shows the breakdown of "call dispositions" for the public opinion survey. The overall response rate of 57% represents the number of completed interviews (n = 802) divided by the number of eligible telephone numbers (n = 1410). Thus, the denominator for this calculation includes incomplete interviews, refusals, language problems, and those who could not be contacted at a verified residential number for whatever reason (e.g., away on vacation, never at home when called, temporarily in hospital, etc.). More than half of the non-response (324 households, or 53% of all non-response) was a result of being unable to contact a respondent in the randomly-selected household. Refusals made up less than half (260 households, or 43%) of the total non-response in this survey.
Table 2-5 displays the breakdown of the final sample by host community and gender of respondents.
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
The public opinion survey was conducted over a period of three weeks (October 6 - 27, 1998) by trained and supervised interviewers and administered through the CATI (Computer-Assisted Telephone Interviewing)7 system installed on a local area network at the PRL. The question text and instructions were presented on a computer screen to the interviewer who asked questions of the respondent over the telephone. Responses were entered directly into the computer. All of the data collection was conducted from a supervised research facility at the University of Alberta between the hours of 9:30 a.m. and 9:00 p.m., seven days a week. Upon making contact, interviewers identified themselves, verified the telephone number, and then asked the screening questions for selecting the respondent. If the interviewers were unsuccessful in establishing contact on their first call, a minimum of fifteen callback attempts were made before declaring a telephone number as "no contact."
Before administering the questionnaire, the interviewer informed the respondents that their participation was voluntary, that their responses would be kept completely confidential, and that they could terminate the interview at any time. After the three-week period of interviewing ended, 10% of the 802 respondents were re-contacted by interviewing supervisors to verify that the interview had been completed with eligible respondents.
At the end of the interviewing period, the survey data were accumulated and formatted for analysis with the SPSS for Windows statistical package. Coding categories for the open-ended questions were constructed by members of the research team after examining the range of responses (recorded verbatim by interviewers) provided by sample members. These responses were then electronically coded by professional coders and added to the SPSS data set. Data cleaning included discrepant value checks and consistency checks.
------------------------