[Continuation 2 of 3:]
Characteristics of effective systems and relationships

Issues in conducting research. Research organizations, of course, are not direct service providers. However, concerns for racial, ethnic, cultural, and linguistic diversity among staff, for understanding of cultural dynamics and the diversity between and within specific populations, and for a commitment to empowerment are all relevant to research as well as service agencies. In a research context, "empowerment" relates to the processes through which researchers determine the focus of research; the specific questions they pose; the ways in which they collect, organize, and analyze data; and the ways in which they attempt to get research outcomes into the hands of those who can use them. Wright and Leung (1993), among others, recommend that "minority communities and persons with disabilities should be involved in the process of planning research, implementing research, and interpreting findings of research" (n.p.). Without "cultural competence" as it is described above, research agencies will find it difficult to secure the involvement of minority groups in any meaningful way.

Some scholars concerned with diversity issues have focused a cultural lens specifically on the processes and assumptions of research. They have found "that data are collected, organized, and communicated within a cultural context and with tools that are also products of the culture" (Metoyer-Duran, 1991, p. 320). As Muntaner, Nieto, and O'Campo (1997) observe, the sociology and history of science show that:

Decisions about basic assumptions guiding research are social phenomena … The scientific community determines in part the acceptability of hypotheses for inquiry, publication, and continued funding through a social process in which certain assumptions are uncritically accepted even in the face of empirical refutation. (p. 263)

The following paragraphs describe some of the issues that have been raised in the literature. These range from overarching, epistemological concerns to specific methodological issues.

Epistemological issues. The examination of cultural beliefs extends not only to methodology, but to the epistemological foundations upon which research methods are constructed. For example, Scheurich and Young (1997) note "a lack of understanding among researchers as to how race is a critically significant epistmological problem in educational research" (p. 4). They point out that "no epistemology is context-free. Yet, all of the epistemologies currently legitimated in education arise exclusively out of the social history of the dominant White race… In other words, our 'logics of inquiry' (Stanfield, 1993) are the social products and practices of the social, historical experiences of Whites" (p. 8). Buchman (1982) discusses the fact that researchers and developers often fail to perceive the influence of their own theories and beliefs on the outcomes of their work; he quotes Nisbett and Ross (1980) regarding "the fallacy of misplaced certainty:"

An important step in reducing people's overconfidence would be taken by leading them to recognize that their interpretations of events, rather than being simple read-outs of data, are inferences that make heavy use of theory. Once one recognizes that the same data would look quite different, and could easily support different beliefs, if those data were viewed from the vantage point of alternative theories, the groundwork for a humbler epistemic stance has been laid. (p. 2)

Costa and Bamossy (1995; see also Taylor and Bogdan, 1984) describe common epistemological errors, describing many studies as "parochial," meaning that they "assume similarity" based on characteristics of one culture and pose that similarity as the norm. Costa and Bamossy also describe "ethnocentric research," in which "one culture's 'universal' theories are imposed on another culture"; in such studies, researchers explore differences, but those differences are examined and explained in reference to norms that are assumed to be universal. In contrast to these approaches, the authors recommend comparative studies, which "search for both similarities and differences" without presuming universal norms or the superiority of one set of cultural characteristics over another (p. 21).

Methodological issues. The literature identifies a host of methodological concerns that can compromise the reliability and validity of research outcomes. As discussed below, these include issues related to research design, sampling and other data collection procedures, and data analysis.

Research design. In spite of some researchers' continuing concerns about the validity and generalizability of qualitative research (Conwal Inc., n.d.), Taylor and Bogdan (1984), as well as others, suggest the strengths of qualitative or ethnographic research designs over the more commonly used positivist methodologies in exploring questions related to minority populations. For example, Harry (1992) in discussing studies of parents of children with disabilities, states that studies of parental attitudes "will be more reliable if they utilize a recursive, open-ended approach" (p. 103).

Hermes (1998) reflects one trend in qualitative research with minority groups. In her report on research with an Ojibwe Indian tribe, she describes an approach of "reciprocity and mutual respect" in which she sought guidance and feedback from the community she was studying at every step in her research, from identifying research questions to collecting, review, and analyzing data. Noting that her relationships with the community "ordered the methods" of her research, she recommends a "guiding principle:" "Be in the community as a member first and a researcher second" (p. 166).

Smart and Smart (1997) describe other research methodologies that appear to hold promise in "unraveling the complex relationship between social conditions and minority status" and assessing strategies for "treatment and prevention of disabilities in minority populations" (p. 12). Citing Adler et al. (1994), they report on promising research designs, including

the use of tree-structured regression and "grade of membership" analyses. Tree-structured regression techniques partition populations into subgroups and then identify different paths to given outcomes. Their strength lies in their ability to analyze numerous, complex, interrelated variables… Likewise, "grade of membership" analysis is able to accommodate larger numbers of variables by developing "ideal descriptions," either theoretically or empirically, of various classes of individuals. This technique becomes more precise with the addition of more and more variables. (p. 13)

Identifying and defining variables. Smart and Smart (1997) and others discuss the common problem of "inadequate definition of research variables" (pp. 32-33), particularly the "validity of racial/ethnic classification" (Duarte & Rice, 1992, p. 13, citing Kumanyika and Golden, 1991). Anderson, Wang, and Houser (1993) cite Hahn (1992), who "believes that the assessment of demographic identity in a society with a culturally diverse and rapidly changing population is extremely difficult." Hahn challenges "the assumption that categories of race and ethnicity are consistently defined when, in fact, terminology and categorization differ from source to source and region to region" (p. 12). Smart and Smart note, for example, that "many researchers have failed to distinguish the various subgroups of the Hispanic population, obscuring important differences between such groups… Different definitions of the terms 'Hispanic,' 'Latino,' and 'Mexican American' have yielded different samples, each with varying socio-demographic characteristics" (p. 33). Also challenged is the assumption "that the racial and ethnic categories used by researchers are adequately understood by those being surveyed" (Anderson, Wang, & Houser, p. 12).

Taylor and Bogdan discuss the problems that can arise when research variables are defined in rigid ways that cannot accommodate the realities of life among the populations studies. As an example, they describe a study that focused on conditions within "single-parent" vs. "dual-parent" homes. In collecting data, researchers found that those two categories failed to capture the variety of living arrangements that characterized most of the homes under study, for example, homes that included extended family members who filled parenting roles. However, because of the study's rigid design, researchers were forced to attempt to fit the data to their pre-existing categories.

Sampling and data collection procedures. As noted in the previous paragraph, sampling procedures can be compromised by poor definition of variables associated with the populations being sampled. In addition, a number of authors note that studies involving nonmainstream populations often rely on inadequate sample size (Anderson, Wang, & Houser, 1993; see also Duarte & Rice, 1992).

Concerns regarding data collection procedures include problems in accessing and obtaining data from a representative sample. Huff (1994), for example, describes the complexities involved in trying to set up a focus group composed of urban African American families: "We have learned that… families… may not have: (1) an address; (2) a telephone; (3) access to banking services; or (4) transportation" (p. 10). Anderson, Wang, and Houser (1993) conducted a needs survey among people with disabilities in Hawaii. Using a mail survey approach, they were unable to obtain an adequate return rate. They concluded that the return rate "may have been affected by the values and traditions within Asian cultures that emphasize the importance of keeping information about the family within its unit" (p. 14).

Researchers' failure to understand and accommodate cultural proprieties can result in the provision of misleading or inaccurate data. Harry (1992) notes that, "given all that is known about power relations between dominant and subordinate cultural groups, it should always be assumed that respondents' initial answers to judgments about the educational system may not be reliable" (p. 103). She quotes Leung (1988), who describes responses to research questions among Asian parents: "Attention and affirmation may only mean courtesy and propriety" (p. 105). Yu (1985, cited by Harry) refers to such patterns of response as "courtesy bias," "which may occur when informants do not yet trust the researcher and when researchers are not aware of the cultural inappropriateness of certain approaches" (Harry, p. 105). Problems also arise in assuring mutual understanding of terms used in survey questions. Taylor and Bogdan caution that "field researchers must start with the premise that words and symbols used in their own worlds may have different meanings in the worlds of their informants" (p. 51).

One major concern in data collection is the use of translators in obtaining data from non-English-speaking subjects. Smart and Smart (1992; see also Harry) describe the problems inherent in the common practice of using friends or other family members to act as interpreters, whether in a research or a service setting:

The use of family and friends as translators violates the right of privacy of the client… When the children of the client are used as translators, the client may be reluctant to fully discuss certain topics or the translator may have a tendency to speak for the client. Use of family members as interpreters alters the family structure. (p. 34)

As noted earlier, the use of interpreters also can result in the loss or misinterpretation of responses. The most effective data collection strategy is for those who conduct interviews, observations, or surveys to speak the respondents' language and to be well oriented to the cultural context in which the research activity is being conducted. The same considerations should apply in settings requiring interpretation for persons who are deaf or hearing impaired.

Data analysis. Two criticisms of data analysis appear most consistently in the literature. The first is misinterpretation based on a lack of cultural understandings or the assumption of normative standards that are, in fact, culturally derived (Duarte & Rice, 1992; Harry 1992). The second is the common failure to account for "within-group differences" when reporting on particular populations. Studies suggest that analyses of racial and ethnic demographics often overemphasize between-group differences and under-emphasize within-group differences, so that differences between groups may be exaggerated, while differences within a specific group may be overlooked (Anderson, Wang, & Houser, 1993; NCDDR, 1996).

---

Top