Consultations between the research team and the funding agency led to the decision to interview all refugee youth (aged 15 to 21) and adult members of refugee family units whenever possible. Consequently, two questionnaires were developed, one for interviews with adult refugees and the second for interviews with refugee youth age 15-21 (see Part 1, Volume 3). A majority of the questions included in the adult version were also included in the youth version, but the latter also contained a number of additional questions about issues unique to youth (e.g., school-work transitions; relationships with parents; educational, career, and family aspirations).
Most of the questions in the two research instruments were developed by the research team specifically for this study of refugee settlement experiences in Alberta. The prior review of research literature on refugee-related topics conducted by the research team identified key themes and central research questions that should be addressed in the surveys. A few questions about attitudes towards multiculturalism were drawn from other sources1. Some of the socio-demographic, health, and attitude measures were modifications of questions used in the Canadian Census, other Statistics Canada national surveys, and province-wide opinion surveys previously conducted by the Population Research Laboratory. In addition, some of the questions in the youth survey were drawn from previous studies of school-work transitions in Canada and elsewhere2. Draft versions of the questionnaire were submitted to CIC personnel, and useful comments about question modifications and additions were received in return.
The questionnaires were devised for a structured interviewing format, with a large number of fixed-response questions. However, to capture the variety of opinions and experiences of refugee respondents, the questionnaires also contained a significant number of open-ended questions. The questionnaires, and the interviewing protocols, were examined and approved by a University of Alberta Research Ethics Committee before data collection began. Before formal interviewing began, a pretest was conducted with a small number of refugees resident in Edmonton. On the basis of this pretest, a few questions were re-worded, some were dropped, and a few new questions were added.
According to CIC records, a total of 9,198 refugees were destined to the seven Alberta host communities in the years 1992 to 1997. The CIC database provided to the research team contained names and addresses for a total of 5,208 government-sponsored and privately-sponsored refugees destined to these seven cities.3 Omitted were individuals who claimed refugee status when arriving in Canada, dependents sponsored by their families (themselves refugees who had arrived earlier), and refugees whose addresses were not available. Thus, the findings from this study of refugee settlement experiences in Alberta can be generalized to the population of privately and government-sponsored Alberta refugees in the 1990s, but not necessarily to refugee claimants or family-sponsored refugees.4
A visual inspection of the names and landing record dates for the 5,208 individuals in the CIC data base was used to estimate the number of families and single individuals in this population (see Table 2-1). A "desired sample," roughly proportional to the number of families and single individuals destined to each of the seven host communities, was then devised. Based on the numbers (of families and singles) in this "desired sample," a systematic sampling strategy (i.e., every nth name or family unit) was used to select a "target sample" of possible respondents in Edmonton, Calgary, Lethbridge, Red Deer and Medicine Hat. All individuals destined to Fort McMurray and Grande Prairie were included in this initial target sample. Across all seven cities, this target sample consisted of 956 individuals.
Using a variety of methods (e.g., direct telephone contact, directory searches, enquiries to cultural organizations and to agencies providing services to refugees), the PRL research team was eventually able to determine where 95% of these 956 individuals were currently living. Assuming that the 47 individuals who could not be located had moved, 60% of the sample members were still living in the city to which they had been originally destined.
Interviews were conducted with 616 of the 909 individuals who could be located (525 adults and 91 youth). This translates into an "interview rate" of 64% for the total target sample. However, because a large minority of the refugees were now living in a wide range of communities across the country, it was not feasible to conduct interviews with all of them. A total of 648 of the 956 members of the target sample were asked if they would participate in the study. Only 32 refused. Thus, the "response rate," for those refugees asked to participate, was 95%.
Seventy-four interviews were conducted with refugees living outside of Alberta (a total of 186 members of the target sample were living outside of the province at the time the survey was completed). Out-of-province interviews were conducted in British Columbia (29), Saskatchewan (1), Ontario (41), Quebec (1), and Nova Scotia (2). Seventy percent of these out-of-province interviews were conducted by telephone. The remainder (n = 22) were face-to-face interviews conducted by members of the research team.
Interviews were conducted during the period July 12-October 21, 1998. The interviewing team consisted of thirteen individuals who were selected on the basis of their interviewing skills, multiple-language proficiency, and experience with refugee issues.
The majority of interviews were completed by ten of these interviewers (seven of whom were refugees themselves). The interviewers participated in a full-day training exercise at the outset, and their efforts were coordinated throughout the study by a full-time interviewing supervisor. Approximately one-third of the interviews were conducted in English, with the remainder being translated into 11 different languages (Bosnian/Serbian/Croatian, Russian, Arabic, Urdu, Spanish, Polish, Somali, Vietnamese, Pashto, Amharic, Farsi). The adult questionnaire was translated into Bosnian, Serbian, Croatian, and Serbo-Croatian versions. For other languages, interviewers translated while conducting the interviews. In addition, response categories (e.g., agree-disagree) were translated into nine languages and then printed on answer sheets used by respondents during the interviews (see Volume 3, Part I, for examples).
Most interviews took place in respondents’ homes. To the extent that it was possible, interviewers attempted to conduct separate, private interviews with respondents, without other household members present. On average, interviews lasted about 80 minutes. All respondents (adults and youth) were paid $20 after the interview was completed, to compensate them for their time and any expenses incurred.
Before the interview began, respondents were again informed about the purpose of the study (most had been given this information when first contacted by telephone) and reminded that their participation was voluntary, that is, they could decline to answer any question, and they could terminate the interview at any time. They were also told that they would not be personally identified in any reports or other publications based on the study. Despite some concerns at the outset about refugees possibly being reluctant to participate in the study, very few difficulties of this type were encountered. Instead, once they had been informed about the purpose of the study, most respondents were eager to participate and indicated that they were pleased that their opinions were being sought.
Table 2-2 profiles the global "region of origin" composition of: (1) the original CIC database of government-sponsored and privately-sponsored refugees destined to Alberta between 1992 and 1997; (2) the original target sample of 956 refugees destined to the seven Alberta cities included in this study; and (3) the final sample of 616 refugees interviewed. While the "region of origin" categories are generally self-explanatory, it should be noted that a large proportion of the African refugees were from Somalia, and a majority of the Southeast Asia refugees were of Vietnamese origin. About one-third of the refugees in the Middle East category were from Afghanistan and a similar proportion were from Iraq. More detailed profiles of each of these "region of origin" groups are presented in Chapter 3.
In the early to mid-1990s, the nation-states that formerly comprised the country of Yugoslavia became a primary source of refugees to Alberta (and to Canada as a whole). Not surprisingly, then, refugees from (former) Yugoslavia comprise a significant proportion of all refugees interviewed in this study. Table 2-2 reveals that 46% of the refugees in the original CIC database were of Yugoslavian origin. However, we also see that 61% of the 616 refugees who were interviewed in this study originally came from Yugoslavia. Thus, the final sample significantly over-represents refugees from this region. Part of the explanation for this over-sampling can be traced back to the point at which the "target sample" was drawn (see the centre columns in Table 2-2). The combination of (1) different proportions of refugees from different regions being destined to different cities and (2) different sampling fractions being used to draw the target sample for each city (see Table 2-1) led, unexpectedly, to an over-representation of Yugoslavian refugees in the target sample (52%). In addition, the fact that former Yugoslavians were more likely to be arriving in Alberta towards the end of the six-year period covered by this study (1992-97) also meant that they were somewhat less likely to have moved out of Alberta, somewhat easier to contact and, ultimately, somewhat more likely to be interviewed.
Table 2-2 also shows that refugees from Central/South America are somewhat over-represented in the final sample, while those who came from the Middle East and Southeast Asia are somewhat under-represented in the final sample. With respect to the Central/South American and Middle East groups, the small variation from the original CIC database occurred at the time the "target sample" was drawn, for the same reasons discussed above. As for the under-representation of Southeast Asian refugees in the final sample, it became apparent during the interviewing process that this group (particularly those from Vietnam) were more difficult to locate, and somewhat less likely to be willing to participate in the study.
Table 2-3 details the number of refugees in the target sample, from each major region of origin, who were destined to each of the seven Alberta cities, along with the number in each sub-category who were interviewed. As noted above, refugees from former Yugoslavia were more likely to be interviewed (a 75% interview rate) while those who came from Southeast Asia were less likely to be interviewed (only 21% of the 96 individuals in the original target sample).
Since refugees from former Yugoslavia are significantly over-represented in the final sample of 616 refugees interviewed, survey results for the total sample could be influenced by this sampling bias, if former Yugoslavians answer differently than do other refugees in the sample. Consequently, in all of the data analyses in the following chapters, comparisons across regions of origin will be conducted in order to check for such a possible sampling bias.
Preparation of the data sets (one each for the adult and youth surveys) began immediately following completion of the interviews. Answers to the fixed-response questions in the questionnaires were keypunched directly into a database designed for analysis with the SPSS for Windows statistical package.5 Coding frameworks for the many "open-ended" questions included in the survey were developed by senior members of the research team after examining the range of verbatim responses provided by refugees. These responses were then electronically coded (again, by senior members of the research team), and added to the database. Subsequent data cleaning included discrepant value checks (ensuring that the data set contained only legitimate response codes) and consistency checks (cross-checking to ensure that only those respondents eligible to answer a specific question had, in fact, answered this question).
------------------------