CHAPTER 1

INTRODUCTION

Science Base

Standards of Scientific Evidence

This Supplement draws on the best available science coming from many disciplines — mental health, health services, history, sociology, and anthropology. The statements made in this Supplement are documented by reference to studies published in the professional literature. Publications are first required to be peer-reviewed by fellow experts to ensure their quality. Quality depends on scientifically rigorous methods of data collection, analysis, and interpretation.

No single study, regardless of the quality of its design, is sufficient by itself to serve as the basis for a conclusion in this Supplement. Findings must be replicated in several studies, and findings must be consistent. The strength or degree of evidence amassed for any conclusion is referred to as the level of evidence.

Assessing the level of evidence is often difficult when findings transcend disciplinary boundaries. Distinct disciplines formulate questions differently. This, in turn, dictates different approaches to designing and conducting research, and the approach often deter-mines how researchers report their findings and conclusions. Even when approaches are similar, investigators in different disciplines frequently employ different terms to describe similar concepts. In seeking to apply scientific standards consistently across the many fields of research reviewed, this Supplement emphasizes two criteria: rigorous methods of inquiry and sufficient data to support major conclusions.

Methodological Issues in Studying Minorities

Because race and ethnicity are hard to define, many scientists discourage the use of these terms in the analysis of disease, unless there is reason to suspect, based on other sources of evidence, that a relationship exists. In general, cause and effect relationships between health status and race and ethnicity have been rare, and when they have been found, they are usually related to lifestyle or other behavioral factors that tend to correlate with racial and ethnic categories. Observed differences between racial and ethnic groups are less likely to be caused by underlying biological differences but rather by factors that co-vary with race, such as income, education, or environment. Even central tendency differences in metabolic rates are overshadowed by the complete overlap in the distribution of metabolic rates across American racial and ethnic groups. Some editors of scientific journals actively discourage presentation of racial and ethnic data unless there is a specific rationale for such analyses.

NIH insists that clinical trials to test treatments include a strongly diverse population of volunteers. This diversity is necessary to ensure that the results of the trials will apply broadly to all populations, including minority groups. According to the theory of clinical trials, it is not necessary to separately analyze subpopulations unless there are empirically based hypotheses about group differences.

Still, the study of mental health in minorities is flourishing, even though researchers face methodological hurdles that make these studies more complex, costly, and difficult to conduct than similar types of investigations in predominantly white communities.

One major consideration is related to the measurement of mental disorders. For example, even when using the DSM system to establish the criteria for different mental disorders and a standardized instrument such as the Composite International Diagnostic Interview Schedule (CIDI) to measure disorders, cultural factors affect how individuals define, evaluate, seek help for, and present their health problems to family members, friends, and service providers. Considering culture in a standardized measure of mental disorders is reliant on at least three types of equivalence: conceptual, scale, and norm. Conceptual equivalence refers to similarities in the meaning of concepts used in assessment: e.g., Do minorities and whites think of well-being, depression, or self-esteem in the same way? Scale equivalence refers to the use of standard formats in questionnaire items that are familiar to all groups. Western-educated people of all groups are familiar with responding to questions that have choices such as “strongly agree,” “agree,” and so on, or a true-false dichotomy. Recent immigrants, particularly individuals who have not been educated in the Western system, may not understand this format. Accordingly, their answers to questions using these response options may not be valid or reliable. Norm equivalence refers to the application of standard norms developed in one sample and used with another group. Because population or sub-population statistics form one standard by which we judge normal and abnormal or high and low functioning, it is important to understand whether the population on which the norms are based is similar to the study group.

Over the past decade, social scientists have used focus groups, ethnographies, and detailed interviews to help modify standardized measures to make them more equivalent for use with racial and ethnic minority groups. Although refining instruments for different racial and ethnic minorities has been made more systematic and efficient, making measures equivalent remains a time-consuming process.

For researchers who use surveys to collect data, a major methodological hurdle is the issue of sampling. Compared with interviewing all members living in a geographic area, sampling is a scientific and cost-effective means to estimate the rates of mental disorder and use of services for a particular group or community. Because ethnic and racial minority groups are relatively rare in most communities, it is difficult to recruit adequate samples for any one particular study. When a study requires large samples of a specific ethnic group, the screening time to locate respondents is quite high. For example, in a study in Los Angeles, nearly 17,000 households were approached to secure a final sample size of 1,747 Chinese American respondents (Takeuchi et al., 1998). If the study design looks for certain sub-groups (e.g., adults, children, and older adults), the cost and time for screening individuals can become even higher.

Another potential obstacle is that racial and ethnic minorities may be reluctant to participate in research studies. For some, like American Indians and African Americans, research raises past breaches of ethics and harm to individuals (Krieger, 1987). For others, like recent Asian or Latino immigrants, participation in research may be a strange concept, and recruitment may be difficult.

In addition to the difficulties of recruiting individual respondents, some racial and ethnic minority communities may resist being part of a research study. Researchers often conduct studies in minority communities because they want their work to have an impact in resolving social problems, guiding policy, or serving as a basis for programs that will improve the quality of life in the community. These investigations can provide communities with needed data to secure resources for new programs, assess interventions that may be useful in the community, or identify high-risk groups. To conduct studies, however, investigators must rely on community cooperation to help identify people and encourage participation. Frequently, an uneasy tension exists between researchers and the communities they study. Community leaders may see researchers as exploitative and divorced from real issues and real-life problems, while researchers view community leaders as compromising research methods and thereby diminishing out-comes, which would have eventually benefited the community. Such tensions can hinder the initiation of research projects in both white and nonwhite communities.