中文摘要：

利用 SAS 软件的 REG 过程步实现了回归分析中异方差的 White 检验，利用加权最小二乘法消除了异方差的影响。通过与残差图法和等级相关系数法的结果比较，该检验效果更精确。利用2013年全国31个省市自治区保险赔付和保费数据进行实证分析。

英文摘要：

Heteroscedasticity phenomenon is tested with white test of REG procedure in SAS,and the weighted least square estimation is used to eliminate heteroscedasticity.The test is compared with both the residual graph and Spearman correlation coefficient method to show more reasonable results. The data of insurance and premium in 2013 is applied to the example analysis.