Lee Sunghee, Satter Delight E, Ponce Ninez A
UCLA Department of Biostatistics, Los Angeles, CA 90025, USA.
Am Indian Alsk Native Ment Health Res. 2009;16(3):1-15. doi: 10.5820/aian.1603.2009.1.
Racial classification is a paramount concern in data collection and analysis for American Indians and Alaska Natives (AI/ANs) and has far-reaching implications in health research. We examine how different racial classifications affect survey weights and consequently change health-related indicators for the AI/AN population in California. Using a very large random population-based sample of AI/ANs, we compared the impact of three weighting strategies on counts and rates of selected health indicators. We found that different weights examined in this study did not change the percentage estimates of health-related variables for AI/ANs, but did influence the population total estimates dramatically. In survey data, different racial classifications and tabulations of AI/ANs could yield discrepancies in weighted estimates for the AI/AN population. Policy makers need to be aware that the choice of racial classification schemes for this racial-political group can generally influence the data they use for decision making.
种族分类是美国印第安人和阿拉斯加原住民（AI/ANs）数据收集和分析中的首要问题，并且在健康研究中具有深远影响。我们研究了不同的种族分类如何影响调查权重，进而改变加利福尼亚州AI/AN人口的健康相关指标。通过使用基于AI/ANs的非常大的随机人群样本，我们比较了三种加权策略对选定健康指标的计数和比率的影响。我们发现，本研究中检验的不同权重并未改变AI/ANs健康相关变量的百分比估计值，但确实对总体估计值产生了显著影响。在调查数据中，AI/ANs的不同种族分类和列表可能会导致AI/AN人口加权估计值出现差异。政策制定者需要意识到，为这个种族政治群体选择种族分类方案通常会影响他们用于决策的数据。
