|
 |
|
|
Minority Representation that Confounds Data Analysis
Your research project will study the impact of a new method for prophylactic treatment to reduce hypertension in elderly nursing home residents. Guidelines for research with human subjects require a minimum threshold level of minority group representation. Earlier studies have shown that the incidence of hypertension varies markedly among racial groups, with black Americans having a higher incidence of hypertension, proportionally, and Americans of Asian decent having a lower incidence. If the study sample were to include minorities, this variation in the incidence of hypertension would confound the analysis and complicate the interpretation of results. Do you:
J1G says: None of the above. Oversample the under-represented groups that need to be represented in the analyses. R3H says: There are several options here. An IRB can always grant an exemption if such an exemption is justified. If the study is sufficiently defined as to the target group, ie a study of risk factors for hypertension in a black population or a study of risk factors for hypertension in a white population, then by definition the study group is defined and non members of the study group would automatically be excluded. The conclusions, of course, are not generalizable. Alternatively, the study group can be skewed to include sufficient numbers for individual sub set as well as total group analysis. Knowingly accepting a biased sample is, under no circumstances, acceptable. Analysis of a larger group, so that minorities can be excluded, is likewise unacceptable, since the target group must be larger than necessary to expose the minimum number of target population members. The excess target group members and all of the members of the group to be excluded would be exposed to potential risks of the study, even if such risks are minimal.
|
