Choosing research topics
Historically, many non-dominant cultural groups have been exploited by researchers with little benefit to themselves.4 As a result, there is often little trust of researchers and little interest in participating in research.5 Focusing on research areas that address self-identified needs of cultural minority groups is one way of establishing trust. Several authors have suggested community-based participatory research (CBPR) as an approach to identifying relevant research topics and promoting participation in research with diverse populations.5–10
According to the US Agency for Healthcare Quality and Research, CBPR is defined as ‘a collaborative research approach that is designed to ensure and establish structures for participation by communities affected by the issue being studied, representatives of organizations, and researchers in all aspects of the research process to improve health and well-being through taking action, including social change’11 (p. 1). CBPR is characterized by an equitable partnership between researchers and community members that entails shared decision making and ownership of the research study and resulting data.4
Daley et al.4 described collaborative efforts in studies with three American Indian (AI) tribes to identify community problems for research. They noted that after problem identification, researchers with related interests were paired with communities to conduct research of interest to both. They cautioned, however, that there was not always a good fit between communities and interest or expertise among members of the research team, and that efforts were sometimes needed to bring in the research expertise in a particular field.
Although the intent of CBPR is to equalize power in collaborative cross-cultural research, that may be difficult to achieve in some cultural groups. De Schweinitz et al.12 found, for example, that despite intensive attempts by researchers, community members in Ghana did not perceive themselves as experiencing full ownership of the project. The authors concluded that, in part, these findings may relate to researchers' emphasis on participation and partnership rather than on values such as equality in decision making, voice and community empowerment. Additional barriers to CBPR proposed by the researchers included lack of a sharply defined organizational structure for CBPR governance and the role of individual citizens vs. representative leaders. Another issue was differences in role assumptions such that community members did not perceive themselves as decision makers but as labour for conducting the study. Finally, environmental challenges related to use of technology for teleconferencing, geographical distance and location of decision-making meetings in academic settings influenced community members' abilities to participate effectively. Perry and Hoffman13 suggested execution of a formal memorandum of understanding between researchers and members of the cultural group to better characterize mutual responsibilities, power and participation in decision making.
Ethical issues in cross-cultural research revolve around processes for the protection of human subjects or ethics assessment, confidentiality, Institutional Review Board (IRB) training and cultural mores regarding informed consent. Each of these issues will be briefly addressed.
Leong and Lyons1 noted that ethical codes, including those related to the conduct of research, reflect the culture and values of the country in which they are developed and that there is potential for conflict between these values and those of US researchers. Similarly, Graffigna et al.14 reported that different approaches to and criteria for ethics assessment of research studies in different countries may complicate cross-cultural research. They also cautioned that such differences may require changes in research approaches given what is approvable (or proscribed) in one locale or discipline vs. another. They noted that such differences are most likely to occur in relation to sampling and recruitment strategies, and that these differences may affect the comparability of data across cultural groups.
Confidentiality is another ethical issue that has cross-cultural implications. For example, Halcomb et al.15 noted that using community members to assist with data collection in focus groups may promote trust and greater sharing of information, but may also impede participation if facilitators or other focus group members are known to participants. They suggested that, although confidentiality of proceedings can and should be reinforced, it cannot be guaranteed, and participants should be made aware of the potential for breach of confidence.
Use of recording, common in much qualitative research, can be another source of concern with respect to confidentiality. Some cultural groups may resist taping and others may be uncomfortable with both taping and note taking if these violate cultural norms common in orally oriented cultures.15,16 Daley et al.4 suggested that audio taping might be more acceptable in some groups than videotaping because it facilitates de-identification of participants. In other cases, researchers may need to rely on memory and immediate debriefing of several research observers immediately after data collection sessions.
Adequate protection of human research subjects requires that all members of the research team be knowledgeable regarding human subjects issues. In the United States, this necessitates assuring that all team members have completed training in the ethical and regulatory issues associated with the protection of human subjects in research. Training programs are often complex and confusing, even for academic researchers, and may be beyond the capacity of community members participating in data collection to comprehend.4 Several approaches may be taken to address this issue. For example, Katz et al.17 chose to provide IRB training for members of their CBPR community advisory board. Daley et al.,4 on the other hand, paired researchers with community members for IRB training. This served a dual purpose of educating community team members regarding human subjects content and provided for greater familiarity between researchers and community members.
Informed consent is another area in which ethical values may conflict with social mores within a particular cultural group. For example, written informed consent may be inappropriate in countries or among groups with no written language or low literacy levels.1 Some authors have suggested use of graphic representations of information in consent materials and audiovisual documentation of consent rather than a signed form. IRBs have sometimes suggested use of a thumbprint in place of a signature, but for some cultural groups with a history of repression, fingerprinting may give rise to unpleasant memories of compulsion and coercion.18
Other concerns regarding informed consent centre on who is empowered to give consent. In some cultures, for example, the father or the male head of the household must give consent for research participation by family members, particularly children, but sometimes wives and other female relatives as well.1 The problem of voluntary participation also surfaces in cultures in which local officials have significant power and may coerce community members into participating.18 Different perceptions of coercion, bribery and cultural gift-giving expectations may also influence voluntary participation. For example, in very low-income groups, monetary reimbursement may be seen as a coercive approach to study participation or may create feelings of resentment among members of the culture not selected for participation.1 Similarly, what might be considered a ‘bribe’ for participating in the dominant US culture might be a cultural expectation of a gift in appreciation of a participant's efforts.18
The final issue with respect to informed consent deals with conveying adequate information on which potential participants can make informed decisions. In some cultural groups, people might voice understanding when they do not, in fact, understand information, to save the researcher from losing face. Similarly, in some cultural groups, discussion of possible risks associated with research participation may be viewed as causing those risks to occur. Similar difficulties with understanding concepts of individual or societal benefit, randomization and blinding, common to informed consent language, may also be encountered.18 Adams et al.18 described a labour-intensive iterative process of developing a culturally appropriate informed consent process for a traditional Tibetan society that addressed many of these challenges and may be informative for readers.
Participant recruitment and retention
Traditionally, members of non-dominant cultural groups have been less likely to participate in research than their dominant culture counterparts. In part, this is due to the lack of trust in researchers discussed earlier, but may also have to do with their relative invisibility within a given society and their preoccupation with other issues related to day-to-day survival. For these reasons, it is often necessary to engage in creative strategies to recruit and retain members of cultural groups to participate in cross-cultural research. This often means specifically seeking out hidden segments of the population to promote participation. For example, Clark et al.20 purposely constituted ethnically and linguistically homogeneous focus groups from specific underrepresented minorities in their assessment of community health needs. Clark and Natipagon-Shah21,22 used snowball sampling in two studies of Thai American women's participation in mammography screening. In another study, my colleagues and I successfully recruited Japanese American study participants in ethnic beauty salons.23 Similarly, other authors have recruited participants in churches and temples, social service agencies serving the population, community events and culturally oriented student organizations.4,21,24,25
Another strategy for recruiting members of the cultural group for study participation involves the use of community liaisons. In Katz et al.'s17 study, the participation of an AI nurse practitioner as part of the research team was integral to subject recruitment. Similarly, Halcomb et al.15 and Fleischhacker et al.26 engaged community leaders and tribal elders to promote community participation in their studies.
Retention of study participants may also be challenging, particularly in longitudinal studies. Members of some non-dominant cultural groups may move frequently, making follow-up contacts difficult. Redwood et al.27 described a system for telephone contact with Alaskan Native (AN) study participants that resulted in exceptionally high retention rates. They conducted telephone calls during evening and weekend hours when participants could be expected to be at home and also used community members to visit participants who did not have telephones in their homes. In addition, they tracked participants through social networks, community agencies and medical records, and developed a computerized tracking system that resulted in an 88% retention rate. They noted that the contacts were somewhat labour intensive but yielded much better retention than in prior similar studies in which an average of nine contacts were required to obtain 75% retention compared to three or fewer contacts required for 49% of their sample and fewer than 10% of participants requiring 10 or more contacts. Given the resulting retention, they felt the effort was worthwhile. It should be noted, however, that their research was conducted in a small community in which community members were well known to each other. Different strategies might need to be employed to promote retention in more geographically dispersed populations.
Data collection strategies
Commonly used data collection strategies in cross-cultural research include qualitative approaches, such as individual and focus group interviews, and quantitative surveys. Challenges and suggestions related to each of these approaches will be discussed here.
A focus group is a group interview technique that generates data on a given topic from several participants at once. Focus groups are particularly useful in gathering data from culturally and linguistically diverse groups.15 Focus groups often provide a safe venue for the expression of opinions that might not be voiced in individual interviews due to perceived power differentials between interviewers and interviewees.4
Focus groups are particularly useful in cultural groups with an oral tradition of problem resolution. For example, a former doctoral student effectively used focus groups to determine the health promotion needs of male Arab Muslims in southern California. Group discussion of problems is an acceptable and common mode of problem solving in this cultural group.
One advantage of focus groups is that the discussion among participants often triggers thoughts and comments that would not have occurred to participants in individual interviews. Within some cultural groups, however, this may not occur. In an early focus group study with AI women, for example, the women treated the focus group as a ‘talking circle’, a traditional approach to information sharing in some tribal cultures. In a talking circle, a ‘talking stick’ is passed from member to member, and only one participant speaks at a time.26 There is no cross-participant discussion or give and take or direct disagreement among participants, which may defeat the purpose of a traditional focus group in identifying differing points of view.
Modified talking circles have been suggested, however, as a means of gaining focus group data from tribal populations.26 Modified talking circles capitalize on the cultural tradition of oral storytelling, but discussion proceeds until group consensus is achieved and the talking stick makes a complete circuit of participants in silence. Unlike traditional talking circles, the facilitator did not participant in the discussion, because the facilitators, although AIs, only represented one tribe and not necessarily the tribe participating. In addition, the modified talking circles emphasized a topical focus rather than the more traditional series of open-ended questions and probes. Fleischhacker et al.26 noted that the use of homogeneous modified talking circles with each of seven separate AI tribes in North Carolina emphasized the sovereignty and unique environment of each tribe with respect to dietary issues. They also reported that the use of talking circles with tribal elders resulted in a better understanding of food-related health issues and promoted support for subsequent research efforts with each tribe.
Focus groups may also have disadvantages with some cultural groups. For example, self-disclosure or reticence is common in some Asian cultures, and people may be reluctant to voice their opinions in a group. Focus groups may also need to be comprised of homogeneous participants (e.g. all male, specific age groups, certain levels of status in the population). Women, in particular, may be uncomfortable in mixed-gender groups in some cultures or in discussing some topics. Focus group members should also be of equal status to minimize the effects of power relationships.15 Researchers should also consider the ramifications of the focus group setting. Many Western cultural groups are comfortable in an open circle, but use of tables or desks may be more comfortable for Asian populations.15
Many researchers have used members of the cultural group to facilitate focus groups.4,11,20,21 This has both advantages and disadvantages. The group member has the advantage of understanding the culture and views the discussion in the context of the culture. Members of small groups may be uncomfortable if the facilitator is well known to them or is a family member. In addition, facilitators may have difficulty refraining from contributing to the discussion and focusing on their facilitator role.17 Other disadvantages include pre-emption of discussion by exceedingly vocal members or the development of a group consensus that may inhibit dissenters from voicing their opinions.15
Some researchers use focus groups in conjunction with individual interviews and other data collection mechanisms. For example Perry and Hoffman13 used focus groups to follow-up on the results of a survey of physical activity among AI youth. Conversely, Clark and Natipagon-Shah21 and Somanchi et al.24 used focus groups to construct culturally appropriate surveys of Thai American women's attitudes and beliefs regarding breast cancer and mammography. Other authors combined focus groups with participant observations in school cafeterias, meetings with school counsellors and administrators, community events and meetings with youth andtheir parents in their homes to address youth risk for suicide.28
Surveys are another commonly used data collection method in cross-cultural research that is fraught with potential perils. Thomas29 noted that surveys and questionnaires are often used because of their ability to obtain systematic comparable data from large groups of people. He also noted, however, that the utility of a particular measure is a function of its transferability from one cultural context to another and stated that ‘lack of appropriate contextualization may lead researchers to incorporate ethnocentric attitudes and perspectives into their studies . . . leading to false conclusions and consequent inappropriate policy implementation’.29 (p. 212) This concern is highlighted by data indicating that information from three US national surveys in the same year yielded different estimates of disparities and rates of uninsurance among the AI/AN population. Discrepancies in findings were related to differences in the wording of questions, their placement within the questionnaire, different periods of recall and differences in data-processing approaches.30 If surveys related to a relatively simple construct like insurance coverage can be misconstrued, how much more likely is it that more complex concepts will be subject to measurement error?
Participants in cross-cultural research differ in their fluency, or even basic knowledge, of the original language of many research instruments, so tools are often translated into the target language of participants.20 Translation is defined as the process of changing some message, whether written or oral, from one language into another.2 Several authors have described detailed processes for ensuring that translated instruments are linguistically comparable.31 Peña,32 however, pointed out that linguistic equivalence is not sufficient to assure the appropriateness of a measure in a different cultural context. Linguistic equivalence reflects similar meanings of items in both original and target languages.
Other elements of equivalence required are functional, cultural and metric equivalence. Functional equivalence means that a particular item on a tool will elicit the same target behaviour in two cultural groups. Functional equivalence can be enhanced by moving away from the original wording of the item to present the concept in a way that is familiar to the target audience. For example, in a test of knowledge of nutrients and their sources, one might ask about foods common to the diet of the particular cultural group. Cultural equivalence means that a given instruction or item will be interpreted in the same way within both cultural contexts. Peña32 gives the example of an item designed to determine when a child first began to talk. In the dominant US culture, parents would report the age at which the child first used simple words like ‘mama’ and ‘dada’. Other cultural groups, however, do not consider that ‘talking’ and would report the age at which the child began using simple sentences. For cultural equivalence, the item would need to be reworded in the second culture to elicit the age at which simple words were used.
Metric equivalence refers to the degree of difficulty of the item in the two languages. For example, an item testing emergency preparedness using the word ‘flashlight’ would be easily understood by most English-speaking Americans, but would be more difficult for their British counterparts who typically use the word ‘torch’. An appropriately translated tool will contain items of a similar level of difficulty in both the original and translated versions, even if the precise wording must be changed to achieve this.32
Henrich et al.33 have pointed out that the large majority of psychological research measures have been developed based on what they refer to as WEIRD samples, people from Western-educated, industrialized, rich and democratic societies. The assumption is often made that these groups adequately represent the world's population. Research, however, indicates that they are actually outliers on a number of different dimensions, including visual perception, spatial reasoning, categorization and inferential deduction. Other areas in which members of these societies often differ from the rest of the population include moral reasoning and general reasoning styles, as well as self-concept and related motivations. In some cases, they indicated that the populations on which many measures have been normed are actually ‘outliers among outliers’ rather than being representative of universal constructs.
Chen34 noted the need to be sure that a given tool is measuring the same construct in a new cultural group as it did in the original group. There may be different definitions and meanings of the same concept in different cultures or concepts in one culture may include a different number of factors than the same concept in another culture. Means of assuring this ‘measurement invariance’ include statistical procedures such as multiple-group confirmatory factor analysis.
Researchers may also need to consider response patterns characteristic of different cultural groups. For example, some cultural groups would tend to respond to Likert-type scales at the extremes of the scale, while others tend toward the midpoint.29,34 Cross-cultural comparisons in such circumstances would be invalid. Several authors have suggested that CBPR can assist in the development of culturally valid translations of instruments,13,35 and others have noted that questionnaires developed on the basis of interviews with group members may be appropriate.29
One last consideration related to self-report surveys reflects the willingness and ability of members of other cultures to complete them, even when appropriately translated into their target language. Some study participants may wish to participate in the intervention, but not to complete the associated paper work, which may be lengthy and tedious. This was the attitude of older Hispanic and black women encountered by a doctoral student exploring the relationship of fatalism to breast cancer screening. Similarly, members of some cultural groups may be unwilling to commit responses to paper due to fear of recriminations. Also, participants from some cultures may have no experience with test taking and so become confused or do poorly as a result of unfamiliar formats.1 Possible solutions include other forms of data collection, such as cognitive interviews and focus groups to determine what people are thinking when answering specific questions,35 observations of actual behaviour,28 and retrieval of information from other data sources, such as family members, teachers and so on.1,28
In addition to the language difficulties implicit in the use of instruments developed in other cultures, language may also influence data analysis strategies, particularly in the interpretation of qualitative data. Just as language influences data collection in surveys, language may complicate analysis of interview transcripts or focus group sessions conducted in another language. Regmi et al.2 noted the need for several researchers to translate and check transcripts for and convey understanding in the linguistic context in which comments are made. They also pointed out the difficulties of assuring cultural equivalence, particularly with respect to colloquialisms in one language that may not translate well into another language. In addition, researchers need to agree on rules for transcribing information, such as how punctuation will be used. The authors suggested that at least two bilingual persons translate research texts into the target language (usually English), with another person to back translate the materials. Translations would then be compared to the original for accuracy and equivalence, and any discrepancies negotiated among the translators. They also recommended literal transcription of all individual and group interviews including pauses, emotional expression and related annotations, followed by translation into the target language.
Fleischhacker et al.26 noted that analysis of talking circle data was less difficult than that of a typical focus group because of smooth transitions among group members and the participation of all circle members. Information from subsequent speakers tends to build on or contrast with prior speakers, making identification of themes easier. They also reported that review of transcripts and discussion with group members strengthened the transparency of the study and the validity of findings, but required time.
Daley et al.4 described a detailed process for analysis of focus group data incorporating a team of researchers, each of whom is assigned responsibility for different levels of coding. They noted that both members of the cultural group and academic researchers should be involved in coding, with the opinions of community members given priority in any disagreements. Incorporation of members of the cultural group in the analysis and development of the code book for qualitative data analysis took time due to the need to train them in coding processes and differences in perspectives; however, their incorporation in code development assured similar perspectives in data coding and interpretation.
Time considerations in cross-cultural research
The last major consideration in cross-cultural research relates to time. As noted earlier, incorporation of community members in CBPR requires time spent in developing rapport and in educating members of the community for participation in research. Community review of findings and approval of publications prior to dissemination may extend the time needed for publication, which may have ramifications for academic researchers faced with tenure and promotion requirements.4 Other aspects of cross-cultural research may also increase the time needed. For example, Adams et al.18 reported that the development of a culturally relevant consent form took 12 to 15 months. Time issues may also affect submission of grant proposals, which usually have specific application deadlines.17
Data collection may also require more time than in other research settings. For example, Fleischhacker et al.,26 reported spending 9 months arranging and completing the modified talking circle process used in their study. Other cultures may also have differing orientations to time than those of researchers. For example, one doctoral student recently reported waiting more than an hour for a participant to arrive for a scheduled interview. She was about to leave when the participant finally arrived. When describing her experience to a co-worker from the same cultural background, the co-worker laughingly commented, ‘Welcome to Samoan time’. Time may also have specific cultural meanings in some ethnic groups. For example, Samoan community advisors suggested that Aitaoto et al. conduct focus groups of 4–5 hours in length because the length of time spent on a topic can indicate its importance.25
Other issues related to time may centre on appropriate times for meetings among community members involved in research studies as well as for participants. For example, the talking circles used by Fleischhacker et al.26 needed to be scheduled at times when the tribes were not involved in competing activities. Similarly research with migrant farm workers should not be scheduled during peak work seasons, and data collection in other settings may need to be scheduled on weekends or in the evening when members of cultural groups are free.15 In addition, CBPR often involves participation by community volunteers with full-time jobs and multiple demands on their time.26 Cross-cultural researchers need to be alert to the meaning of time within a particular cultural group as well as the time constraints under which community partners and study participants function.