Results for Imputation Strategies 1, 2, and 3 at the Census tract level stratified by urban/rural geographic region.

Presented are the percentage of times the 95% multiple imputation intervals contained the true total case enumerations and the width of the imputation based intervals (in units of number of cases), both averaged across the tracts within each region. Results based on the geocoded Maryland Prostate cancer data split into the experimental geocodes and experimental nongeocodes subsets for evaluation.

本文呈现两项统计结果：一是95%多重插补区间（multiple imputation intervals）覆盖真实总病例枚举数的频次占比，二是基于插补构建的区间宽度（单位为病例数）；两项指标均在各区域内的普查片区（tracts）上计算平均值。上述结果基于已地理编码（geocoded）的马里兰州前列腺癌数据集生成，该数据集被划分为实验地理编码子集与实验非地理编码子集以开展评估。