Results
The sampling response rate for each group varied. For the Kurds in Christchurch 65% of eligible people who were approached participated, while snowballing within the Christchurch Afghan community achieved an 85% response. For both Kurdish and Afghani groups in Perth the response rate was only about 40%, even with snowball endorsement.
The overall sample of 193 participants (Christchurch n = 98, Perth n = 95) reflected a good spread of participants, with a range of opinions and experiences (Table
3). There was a good gender balance, age distribution, family size (1-10), time spent as a refugee (< 1-27 years), and time since resettlement (< 1-20 years). Education level was determined by the number of years of schooling with less than one year coded as none/minimal, 1-6 years as primary, 7-13 years as secondary, and the remainder classified as tertiary level.
Table 3
Summary of participant demographics
Refugee community
| | | | | | | |
Afghan | 90 (47) | 90 | | 49 | 41 | χ2(1,193) = 0.91 (location*) | 0.341 |
Kurdish | 103 (53) | | 103 | 49 | 54 | | |
Gender
| | | | | | | |
Male | 100 (52) | 45 | 55 | 59 | 41 | χ2(1,193) = 0.22 (community**) | 0.637 |
Female | 93 (48) | 45 | 48 | 39 | 54 | χ2(1,193) = 5.6 (location) | 0.018 |
Age
| | | | | | | |
18-19 years | 17 (9) | 4 | 13 | 11 | 6 | | |
20-29 years | 62 (32) | 35 | 27 | 29 | 33 | | |
30-39 years | 56 (29) | 25 | 31 | 26 | 30 | | |
40-49 years | 33 (17) | 11 | 22 | 19 | 14 | | |
50-59 years | 17 (9) | 10 | 7 | 8 | 9 | | |
60 years & over | 8 (4) | 5 | 3 | 5 | 3 | | |
Marital status
| | | | | | | |
Married | 125 (65) | 59 | 66 | 59 | 66 | χ2(1,191) = 2.51 (community) | 0.474 |
Never married | 55 (28) | 24 | 31 | 31 | 24 | χ2(1,191) = 2.06 (location) | 0.561 |
Previously married | 11 (6) | 5 | 6 | 7 | 4 | | |
Religion
| | | | | | | |
Muslim | 180 (93) | 88 | 92 | 88 | 92 | | |
Non-Muslim | 8 (4) | 1 | 7 | 7 | 1 | | |
Education
| | | | | | | |
None | 14 (7) | 5 | 9 | 9 | 5 | χ2(1,193) = 10.26(community) | 0.017 |
Primary | 28 (15) | 9 | 19 | 25 | 3 | χ2(1,193) = 22.70 (location) | 0.000 |
Secondary | 100 (52) | 43 | 57 | 45 | 55 | | |
Vocational/university | 51 (26) | 33 | 18 | 19 | 32 | | |
Continuous Variables
|
Range
| | | | | | |
Time as refugee yrs (Median) | 0-27 years | 4.0 | 8.0 | 15.0 | 2.0 | U = 3451.5, Z = -2.74 (community) | 0.006 |
| | | | | | U = 2159.0, Z = -6.24 (location) | 0.000 |
Time since resettlement yrs (Median) | 0.5-20 years | 6.0 | 7.0 | 4.0 | 11.0 | U = 3675.0, Z = -2.38 (community) | 0.017 |
| | | | | | U = 1459.5, Z = -8.22 (location) | 0.000 |
Mean household size (SD) | 1-10 people | 5.0 (1.8) | 5.4 (2.0) | 5.5 (1.9) | 5.0 (1.9) | | |
No significant differences in participant numbers between groups were noted for the total number of Afghan and Kurdish participants, the total number of participants in Christchurch and Perth, marital status, and household size. However, significant gender differences by location were found, with more males in Christchurch (60%) than Perth (43%). Differences by resettlement location and between refugee communities were also observed for the total time spent in a refugee situation (Kurds longer than Afghans, and those in Christchurch longer than people in Perth), the time since resettlement (participants in Perth settled longer), and education level of participants (those in Perth reported more years of schooling overall than the group in Christchurch).
Representativeness of the sample
Lack of a clear sampling frame [
13,
25,
26], a widely acknowledged problem with refugee research, and limitations with Census data [
8,
13,
28] hinder attempts to accurately gauge representativeness. However, to assess the likely representativeness of our sample, aggregated Afghan and Kurdish data was compared with 2006 Census data for Western Australian (WA) residents born in Afghanistan, Iran and Iraq (Table
4) [
29‐
31]. Appropriate Census data from New Zealand was not available for comparison, although anecdotal reports from within the communities suggest that our sample reflected local demographics. With such hidden populations, it has been suggested that construction of a 'tentative map' of community demographics developed in consultation with professionals working with the target groups or community members may be helpful in order to judge representativeness [
26]. In our case, this was done informally through consultation and discussion with community members. Due to the invisibility of Kurdish ethnicity in immigration statistics, Iraqi and Iranian census data was used as a compromise substitute, although significant limitations exist with this. In particular, the Kurdish geographic region spreads across Iran, Iraq, Turkey, and Syria, with small populations elsewhere in the Middle East, and it is an individual's country of birth that is recorded in government statistics. As the Kurds comprise minority groups in these countries, a direct comparison with natal population demographics is likely to be misleading. However, as shown in Table
4 our sample did achieve a good gender and age balance when compared with the ethnic populations in WA. Religion was predominantly Muslim in our sample, which is consistent for people of Afghan and Kurdish background. The Iran and Iraq-born groups in WA also include a significant number of Baha'is, Armenian apostolic adherents and Catholics who sought refuge from religious persecution as minorities in the 1980 s [
31], which explains the disparity with this variable. Overall, our Afghan sample was slightly better educated (59% with some form of tertiary qualification, compared with 34% of Afghans in WA), and both groups were settled longer (almost half arrived before 1996, compared with about one third of the respective populations). Our sample also appears to speak better English; however, this result is distorted by the definition of speaking English. In the Census, participants are asked to rate their English ability as speaking English very well, well, not well or not at all; only those who speak it well or very well were included in the table data. In comparison, we dichotomised responses into functional/no functional English to assess levels of English literacy, so our results will appear positively skewed for this variable.
Table 4
Comparison of total sample demographics with 2006 Census data for Western Australia
Place of birth
| Iran | 4 | | 39 | | |
| Iraq | | | 59 | | |
| Afghanistan | 78 | | | | |
| India | 4 | | | | |
| Pakistan | 3 | | | | |
| Turkey | | | 3 | | |
| Not stated | 1 | | 2 | | |
Gender
| Male | 50% | 54% | 53% | 53% | 52% |
| Female | 50% | 46% | 47% | 47% | 48% |
Median age
| | 20-29 | 28.9 years | 30-39 | 35.7 yrs | 40.4 yrs |
Religion*
| Muslim | 98% | 95% | 89% | 31% | 32% |
Speaks English **
| | 76% | 68% | 95% | 71% | 81% |
Post school
qualifications ***
| | 59% | 34% | 33% | 33% | 59% |
Arrival pre-1996
| | 46% | 27% | 52% | 33% | 55% |
Discussion
Numerous methodological and ethical challenges arise when conducting research studies with refugee or migrant groups. One fundamental decision is whether to adopt a quantitative or qualitative approach. Both methodologies have their own merits, depending on the research focus and desired outcomes and can be distinguished primarily by the type of data obtained (text based or numeric), the underlying logic employed (inductive or deductive), method of analysis used (interpretive or statistical) and the presumed underlying paradigm (positivist/rationalistic or interpretive/critical/naturalistic) [
32]. Increasingly, researchers are adopting mixed method approaches, employing innovative strategies to combine methods that attempt to both generalise results to a wider population, while also generating in-depth understanding of individual cases [
32,
33]. Mixed method techniques can be used to expand the scope of a study, to corroborate data, provide deeper insights, or aid ongoing development of a project. In our study, quantitative instrument-based data was considered to provide complementary data (rather than for statistical inference) to corroborate interview findings and allow idiographic generalisations about study participants rather than nomothetic generalisation about the populations as a whole [
33]. Kessler-10 scores, for example, could also be used for qualitative profiling, by allowing further in-depth qualitative analysis for participants scoring within the high or very high risk of psychological distress range.
One of the main methodological concerns with mixing methods arises when selecting the type of sampling strategy to be used. Probability techniques, such as simple random, stratified (sampled to meet fixed quotas based on previously defined variables), systematic random or cluster sampling methods underlie quantitative methodologies, allowing statistical inferences based on generalisations from the sample to a wider population to be drawn [
33‐
35]. These are also referred to as descending methodologies because of the focus on the general population from which the sample is drawn. In contrast, purposeful or non-probability sampling is an ascending method, working up from individual cases to draw conclusions and generate idiographic knowledge [
33]. Common techniques include convenience sampling, where cases are chosen because they are readily identified or available, quota sampling (similar to stratified sampling but cases are not randomly selected), random purposeful (selection from large pool of info-rich cases) or stratified purposeful methods (based on a pre-set quota of info-rich cases) and snowball sampling (mainly used to identify rare cases, where sampling depends on referrals from existing cases to generate potential participants) [
33,
35]. For many studies the design and sampling technique will be dictated by individual research goals, however the unique set of challenges encountered when working with hidden or hard to reach populations, such as refugee groups, necessitates a more pragmatic approach. In practice, most research studies with forced migrants employ some form of non-probability sampling, adopt an ethnographic community participatory approach [
8,
14], or in some cases are able to utilize pre-existing data sources, such as Census or Immigration statistics [
13,
28] although these have only limited capacity for analysis of specific issues or identification of former refugees.
A 'hidden population' refers 'to a subset of the general population whose membership is not readily distinguished or enumerated based on existing knowledge and/or sampling capabilities' [
15]. The lack of clear sampling frames means that snowball sampling may be the only feasible way of locating potential participants [
26,
27], despite concerns around selection bias [
13,
25‐
27]. It is conceivable that sampling frames may exist for some sub-groups of refugees, for example those currently within a resettlement program [
13], however for those settled longer, individuals who may have internally migrated, asylum seekers or families arriving under reunification programs, few, if any, records will be available. Many minority groups may be socially invisible, effectively hidden within existing population statistics, which is a particular issue for those of Kurdish ethnicity who are normally categorised by their country of birth (mostly Iraq, Iran, Syria or Turkey), and presents a particular challenge for recruitment. It is not possible to obtain accurate immigration or census data for these groups, so utilisation of nationality-based databases for random selection is not feasible.
Sampling challenges are of fundamental concern when the entire validity of a research study may be questioned depending on the method selected. Due to sampling difficulties, refugee-focussed research projects often either utilise existing epidemiological databases, are conducted on a large scale through governmental or health organisations where attempts at randomisation with multiple ethnic groups are feasible, or employ purposive sampling on a smaller scale with a qualitative focus. This presents a dilemma when attempting to obtain comparable data between groups or locations; either some compromise is required to access and recruit participants, or ethnic minority groups risk exclusion from comparative research. Alternatively, studies may rely solely on qualitative content, which although offering valuable insight into individual experiences, only tells part of the story and potentially limits monitoring of resettlement outcomes. For this reason, a mixed methods approach, incorporating both qualitative and quantitative dimensions was our preferred option, although as identified during our review of previous studies, there are significant limitations with obtaining representative refugee samples with groups from our region of interest for quantitative studies. For us, and most people with an interest in the health of refugee populations, there is need reach a pragmatic compromise between representativeness and logistic feasibility; without this, there would be a lack of evidence on the health and needs of such vulnerable populations.
Both Christchurch and Perth provide dedicated settlement services and targeted support for refugees, but accurate demographic data of specific communities is scant. Obtaining statistically representative samples of such socially invisible groups is known to be problematic [
13,
25], given the limited size of communities, their invisibility in national data sets, the target participants' concern about research motives, power differentials between participants and the researchers, as well as difficulties with access and trust [
26,
36]. For refugees, some reticence may also be related to pre-migration experiences, so extra care and sensitivity around the establishment of relationships is necessary. When four distinct communities are involved, these issues are multiplied. It was important to develop good relationships with leaders and high profile members of each community, to build rapport, establish our credentials, discuss research objectives and plan access strategies [
14]. This initial preparatory phase, during which we also needed to address language concerns and instrument selection, had an inevitable impact on the time required for the research. The method described here utilised a combination of strategies; the entire Kurdish group in Christchurch was contacted, as it was small and well defined, however the size and difficulties with access to the other three groups necessitated a different approach. By employing a snowball method, we were able to access socially invisible individuals and provide reassurance about our objectives, while at the same time attempting to obtain a cross-sectional representation of the population by using multiple contacts from different groups and backgrounds. The sample size was deliberately large to enhance heterogeneity, and we had multiple, small snowball chains with initial selection guided by an informal quota [
13].
Although limitations in the generalisability of this approach are acknowledged, one strength of this method was in the number of women recruited; especially those with limited education who are sometimes overlooked in research studies and would often decline to participate if approached directly. Overall we achieved a good gender balance (male 52%, female 48%), although when broken down by location the proportion varied amongst Afghan participants, with females being 35% in Christchurch, compared with 68% in Perth. Some variation may be explained by our initial recruitment strategies; as one Christchurch entry point was a male Afghan soccer team whose female relatives were generally not interested, while in Perth we were able to directly approach women through support groups and ESOL classes, achieving a good response rate. Differences in social demographics may also play a part, with many Afghan men who had settled longer in Australia too busy working to participate. In comparison, women who were often at home or attending classes during the day welcomed the opportunity to express their views as it provided them a sense of purpose and empowerment [
37]. In many cases, the men initially arrived as asylum seekers, subsequently being joined by family members under reunification programs; this accounted for some gender disparity in settlement time as well as English language ability. We found more variation among different Afghan ethnic groups in Christchurch, with gender balance among Hazaras but fewer women of other ethnicities chose to participate. Interestingly, recently arrived women in Christchurch (1-2 years resettled) were more motivated to participate than those settled longer in both locations.
The groups in Christchurch had all arrived within the previous ten years, whereas over 40% of the Afghans and 60% of Kurds in Perth had been settled between 11-20 years. It was notable that the attitudes and behaviour of many of those resettled longer were more suspicious of the research objectives and less hospitable. Perhaps this was a reaction to Australian society where refugee issues are hotly debated in the media and research on these topics more common than in New Zealand, or maybe they no longer thought of themselves as refugees. Whether it was a result of research fatigue [
27], a response to public attitudes towards refugees and migrants, or simple apathy, but the opportunity to visit people in their own homes to establish relationships was not really encouraged in Perth. Many Australian participants preferred minimal contact, to meet in a neutral place, or simply to complete questionnaires and return them by mail. The biggest challenge, particularly in Australia, was generating initial interest to overcome apathy and apparent suspicion, with the key feature being identification of enthusiastic people for snowball initiation. Without the chain referral endorsement of snowball sampling, it is unlikely that sufficient interest would have been generated to recruit sufficiently for a viable study. Bloch also noted differences in attitudes between similar refugees in the United Kingdom and South Africa, suggesting that larger numbers of refugees and asylum seekers and more political activists in the UK lead to increasingly suspicious attitudes to research motives [
13]. Our Australian experience was a marked contrast to the support and hospitality received in Christchurch where it was not uncommon for participants to want to entertain with lavish meals once the interview was completed. Although this may be partly explained by the fact that some pre-existing relationships existed, the majority of New Zealand participants were total strangers prior to recruitment. For many, especially more recently arrived women, it was the first time their views and opinions had been sought, so participation proved a novel and rewarding experience which was often fuelled by altruistic motives to help other refugees through their own insight and experience [
8,
37].
Obtaining access to the different communities was a challenge. Overall, it was easier with the groups in Christchurch because we had a better understanding of community dynamics and already established connections with key individuals. Building personal relationships is essential, and an appreciation of customs and social mores fundamental to establishing researcher credentials and acceptability within the group. Taking the time to drink tea and connect on a personal level prior to commencing the formal part of the interview greatly influenced the quality of data obtained. We must have consumed hundreds of cups of tea during the data collection phase, but the opportunity to interact and chat informally provided valuable insights and enhanced our standing overall. The fact that the interviewers were Muslim also improved our respectability, especially for many older people, and helped overcome barriers despite the language issues. Maintaining Islamic proprieties was also important; ensuring interviewers were of an appropriate gender and having an accompanying male relative escort the lead interviewer when necessary, was appreciated and commented on favourably by many participants.
The selection of groups for cross-country studies poses another dilemma, as evidenced by the demographic balance of our sample. Although the entire NZ sample arrived during the previous 10 years, more than half the Australian group were settled longer than this time, in some cases up to twenty years, and this could potentially limit the relevance of some comparative analyses. Variations in migration patterns and conflict situations, national resettlement quotas determined annually in consultation with UNHCR requirements, the presence of existing communities, and many other considerations, result in different refugee groups being accepted for resettlement at different times in different countries. The question arises whether it is better to compare groups based on temporal considerations or ethnic similarity. Because the majority of Afghans and Kurds are Muslims, many are from traditional backgrounds and include members who suffered ongoing ethnic and religious persecution as minority groups; we considered comparison between locations required that the study groups be ethnically similar.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
CS-H conceived the study, participated in its design, co-ordination and data collection, and drafted the manuscript. ST participated in the design of the study and helped draft the manuscript. Both authors read and approved the final manuscript.