Jepson C, Kessler L G, Portnoy B, Gibbs T
National Cancer Institute, Bethesda, MD.
Am J Public Health. 1991 Apr;81(4):501-4. doi: 10.2105/ajph.81.4.501.
Data from the 1987 National Health Interview Survey Cancer Control Supplement were used to estimate multivariate logistic regression models of diet change, mammography utilization, stool blood test utilization, and smoking. Predictor variables included race, sex, age, income, dietary concerns, and four knowledge-related variables: education and three measures of cancer prevention knowledge. When knowledge variables were included in the models, race was not a significant predictor of behavior, with one exception: among women, Blacks were found to smoke less than Whites.
1987年全国健康访谈调查癌症控制补充调查的数据被用于估计饮食变化、乳房X光检查利用率、粪便潜血检查利用率和吸烟情况的多变量逻辑回归模型。预测变量包括种族、性别、年龄、收入、饮食关注点以及四个与知识相关的变量:教育程度和三项癌症预防知识指标。当模型中纳入知识变量时,种族并非行为的显著预测因素,但有一个例外:在女性中,发现黑人吸烟比白人少。